WiMAX Networks: Techno-Economic Vision and Challenges 9048187516, 9789048187515

Citation preview

WiMAX Networks

.

Ramjee Prasad

l

Fernando J. Velez

WiMAX Networks Techno-Economic Vision and Challenges

Ramjee Prasad Center for TeleInFrastruktur (CTIF) Aalborg University Inst. Electronic Systems Niels Jernes Vej 12 9220 Aalborg Denmark [email protected]

Fernando J. Velez Universidade da Beira Interior Instituto de Telecomunicac¸o˜es Depto. Engenharia Electromecaˆnica Calcada Fonte do Lameiro 6201-001 Covilha Portugal [email protected] King’s College London Centre for Telecommunications Research Strand London WC2R 2LS UK

ISBN 978-90-481-8751-5 e-ISBN 978-90-481-8752-2 DOI 10.1007/978-90-481-8752-2 Springer Dordrecht Heidelberg London New York Library of Congress Control Number: 2010928698 # Springer Science+Business Media B.V. 2010 No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Cover design: eStudio Calamar S.L., F. Steinen-Broo, Pau/Girona, Spain Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)

To our families and to our students

.

Preface

Attraction and aversion of the senses to their corresponding sense objects is unavoidable. One should not be controlled by them; since they are obstacles in one’s path. —The Bhagvad-Gita (3.34)

Worldwide Interoperability for Microwave Access (WIMAX) is a technology covering broad range of topics. To the best of our knowledge, WiMAX Networks is a first book that deals with all the relevant areas shown in Fig. 1 namely, quality of service (QoS), security, mobility, radio resource management (RRM), multiple input multiple output (MIMO) antenna, planning, and cost/revenue optimization, medium access control (MAC) layer, physical layer, network layer, and so on. Chapter 1 introduces the WiMAX and locates its place among the existing and future wireless systems. Aspects of OFDM and OFDMA WiMAX physical layer is well covered in Chapter 2. Chapter 3 covers link layer issues, having its main focus on Medium Access Control (MAC) Layer. The term quality of service (QoS) is clearly explained in Chapter 4, considering the WiMAX into account. Security is a primary subject for WiMAX and for secure communications, privacy and confidentiality are fundamental issues. Chapter 5 has taken care of this important subject. vii

viii

Preface

WiMAX Networks

Fig. 1 Tree structure of the book

Chapter 6 presents mobility architecture with integrated QoS support and the proposed architecture can accommodate different wired and wireless technologies. The radio resource management (RRM) in OFDMA based cellular networks such as WiMAX is addressed in Chapter 7. Four different sub-carriers allocation algorithm with low complexity are evaluated for WiMAX cellular systems. Chapter 8 first discusses the propagation models and then introduces the cellular planning in the context of WiMAX. A model to compute the support physical throughput is proposed for WiMAX in Chapter 9 as a function of the achievable carrier-to-noise-plus-interference ratio (CNIR). Chapter 10 first introduces general aspects about the business models for WiMAX and then address the cost/revenue optimization for these networks, for cellular configuration with and without relays.

Preface

ix

Multiple Input and Multiple Output (MIMO) technology options for the WiMAX has been discussed in Chapter 11. Finally, Chapter 12 concludes the WiMAX Networks by comparing WiMAX with other wireless standards and highlights its potential. We would greatly appreciate if readers would provide extra effort in improving of the quality of the book by pointing out any errors. We strongly believe nothing is errorless. Ramjee Prasad Fernando J. Velez

.

About Authors

Ramjee Prasad is currently Professor and Director of Center for Teleinfrastruktur (CTIF), and holds the chair of wireless information and multimedia communications. He was coordinator of European Commission Sixth Framework Integrated Project MAGNET (My personal Adaptive Global NET) and MAGNET Beyond. He was involved in the European ACTS project FRAMES (Future Radio Wideband Multiple Access Systems) as a project leader in Delft University. He was also project leader of several international, industrially funded projects of Technology. He has published over 700 technical papers, contributed to several books, and has authored, co-authored, and edited over 25 books. His latest book is “Introduction to Ultra Wideband for Wireless Communications”. Prof. Prasad has served as a member of the advisory and program committees of several IEEE international conferences. He has also presented keynote speeches, and delivered papers and tutorials on WPMC at various universities, technical institutions, and IEEE conferences. He was also a member of the European cooperation in the scientific and technical research (COST-231) project dealing with the evolution of land mobile radio (including personal) communications as an expert for The Netherlands, and he was a member of the COST-259 project. He was the founder and chairman of the IEEE Vehicular Technology/Communications Society Joint Chapter, Benelux Section, and is now the honorary chairman. In addition, Prof. Prasad is the founder of the IEEE Symposium on Communications and Vehicular Technology (SCVT) in the Benelux, and he was the symposium chairman of SCVT’93. Presently, he is the Chairman of IEEE Vehicular Technology/ Communications/Information Theory/Aerospace and Electronics Systems/Society Joint Chapter, Denmark Section. In addition, Prof. Prasad is the coordinating editor and editor-in-chief of the Springer International Journal on Wireless Personal Communications. He was the technical program chairman of the PIMRC’94 International Symposium held in The Hague, The Netherlands, from September 19 to 23, 1994 and also of the Third Communication Theory Mini-Conference in Conjunction with GLOBECOM’94, held in San Francisco, California, from November 27 to 30, 1994. He was the

xi

xii

About Authors

conference chairman of the fiftieth IEEE Vehicular Technology Conference and the steering committee chairman of the second International Symposium WPMC, both held in Amsterdam, The Netherlands, from September 19 to 23, 1999. He was the general chairman of WPMC’01 which was held in Aalborg, Denmark, from September 9 to 12, 2001, and of the first International Wireless Summit (IWS 2005) held also in Aalborg, Denmark on September 17–22, 2005. He was the General Chair of the First International Conference on Wireless Communication, Vehicular Technology, Information Theory and Aerospace & Electronic Systems Technology (Wireless VITAE) held on May 17–20, 2009 in Aalborg. Prof. Prasad was also the founding chairman of the European Center of Excellence in Telecommunications, known as HERMES and now he is the honorary chairman. He is a fellow of IEEE, a fellow of IETE, a fellow of IET, a member of The Netherlands Electronics and Radio Society (NERG), and a member of IDA (Engineering Society in Denmark). Prof. Prasad is advisor to several multinational companies. He has received several international awards; one of this is the “Telenor Nordic 2005 Research Prize” (website: http://www.telenor.no/om/). Fernando J. Velez received the Licenciado, M.Sc. and Ph.D. degrees in Electrical and Computer Engineering from Instituto Superior Te´cnico, Technical University of Lisbon in 1993, 1996 and 2001, respectively. Since 1995 he has been with the Department of Electromechanical Engineering of University of Beira Interior, Covilha˜, Portugal, where he is Assistant Professor. He is also researcher at Instituto de Telecomunicac¸o˜es, Lisbon. He made or makes part of the teams of RACE/MBS, ACTS/SAMBA, COST 259, COST 273, COST 290, ISTSEACORN, IST-UNITE, and COST 2100 European projects, he participated in SEMENTE and SMART-CLOTHING Portuguese projects, and he was the coordinator of four Portuguese projects: SAMURAI, MULTIPLAN, CROSSNET, and MobileMAN. Prof. Velez has authored five book chapters, around 75 papers and communications in international journals and conferences, plus 25 in national conferences, and is a senior member of IEEE and Ordem dos Engenheiros (EUREL), and a member of IET and IAENG. His main research areas are cellular planning tools, traffic from mobility, simulation of wireless networks, cross-layer design, inter-working, multiservice traffic and cost/revenue performance of advanced mobile communication systems. Susana Sargento (Ph.D. in 2003 in Electrical Engineering) joined the Department of Computer Science of the University of Porto in September 2002, and is in the University of Aveiro and the Institute of Telecommunications since February 2004. She is also a Guest Faculty of the Department of Electrical and Computer Engineering from Carnegie Mellon University, USA, since August 2008. Faculty of ICTI: Information & Communication Technologies Institute. She has been involved in several national and European projects, taking leaderships of several activities in the projects, such as the QoS and ad-hoc networks integration activity in the FP6 ISTDaidalos Project. She is currently involved in several FP7 projects (4WARD, EuroNF, C-Cast), national projects, and CMU Portugal projects (DRIVE-IN with the Carnegie Melon University). She has been organizing several conferences, such as NGI’09 and IEEE ISCC’07. She has been also in the Technical Program Committee

About Authors

xiii

of several international conferences and workshops (more than 20 in 2009), such as ACM MobiCom 2009 Workshop CHANTS and IEEE ISCC (2005–2010). She has also been a reviewer of numerous international conferences and journals, such as IEEE Wireless Communications, IEEE Networks, IEEE Communications, Telecommunications Systems Journal, IEEE Globecom, IEEE ISCC and IEEE VTC. Prof. Sargento is also Area Editor of International Journal of Communication Networks and Information Security. Her main research interests are in the areas of next generation and heterogeneous networks, infrastructure, mesh and ad-hoc networks, more specifically in QoS, mobility, routing, multicast, self-management and cognitive issues, where she has published more than 150 scientific papers. Pedro Neves received his B.S. and M.S. degrees in Electronics and Telecommunications Engineering from the University of Aveiro, Portugal, in 2003 and 2006 respectively. From 2003 to 2006 he joined the Telecommunications Institute, Portugal, and participated in the DAIDALOS-I and DAIDALOS-II European funded projects. Since 2006 he joined Portugal Telecom Inovac¸a˜o, and he is involved in the WEIRD and HURRICANE European funded projects. Furthermore, since 2007 he is also pursuing a Ph.D. in Telecommunications and Informatics Engineering at the University of Aveiro, Portugal. Eng. Neves has been involved in six, as well as more than 25 scientific papers in major journals and international conferences. His research interests are focused on broadband wireless access technologies, mobility and QoS management in all-IP heterogeneous environments, multicast and broadcast services, as well as mesh networks. Ricardo Matos is a Ph.D. student at University of Aveiro. He concluded in July 2008 his Integrated M.Sc. in Electronics and Telecommunications Engineering from the Department of Electronics, Telecommunications and Informatics of University of Aveiro. His master’s Thesis was titled “Mobility Support in WiMAX Networks”. Since September 2008 he joined Institute of Telecommunications (located in University of Aveiro Campus) as a researcher and Ph.D. student. His current research interests are related with future Internet architectures, virtualization techniques, wireless mesh networking, self-management and context-awareness. Marilia Curado is an Assistant Professor at the Department of Informatics Engineering of the University of Coimbra, Portugal, from where she got a Ph.D. in Informatics Engineering on the subject of Quality of Service Routing, in 2005. Her research interests are Quality of Service, Mobility, Routing, and Resilience. Prof. Curado has participated in several national projects, in Networks of Excellence from IST FP5 and FP6, in the IST FP6 Integrated Projects, EuQoS and WEIRD, and on ICT FP7 STREPs MICIE and GINSENG. Bruno Sousa studied computer science at the Polytechnic Institute of Leiria (B.Sc. 2005) and University of Coimbra, Portugal (M.Sc. 2008). He has participated in the IST FP6 Integrated Project WEIRD – WiMAX Extensions to Isolated Reasearch Data Networks. He is now enrolled at the University of Coimbra as a Ph.D. student, with interest areas including mobile computing; multihoming; WiMAX and simulation. Kostas Pentikousis studied computer science at the Aristotle University of Thessaloniki, Greece (B.Sc. 1996) and the State University of New York at Stony

xiv

About Authors

Brook, USA (M.Sc. 2000, Ph.D. 2004). He has published internationally in several areas, such as, network architecture and design, mobile computing, applications and services, local and wide-area networks, and simulation and modeling. Dr. Pentikousis is a Senior Research Scientist at VTT Technical Research Centre of Finland. Giada Landi received the Laurea degree “cum laude” in Telecommunication Engineering from the University of Pisa, Italy, in 2005. Currently she is R&D solutions architect at Nextworks S.r.l.. Her main research activities concern the design, development and integration of Control Plane in network architectures supporting heterogeneous wireless access technologies, with main focus on Quality of Service, Admission Control, AAA and inter-technology mobility. She participated in the IST FP6 WEIRD project and the Artemis SOFIA project. Ljupcˇo Jorguseski currently works as senior scientist/innovator on the optimization and performance estimation of wireless access networks. He has got his Dipl. Ing. degree in 1996 from the Faculty of Electrical Engineering at the university “Sts. Cyril and Methodius”, Skopje, Republic of Macedonia. From 1997 to 1999 he worked as applied scientist at TU Delft, Delft, The Netherlands on the European FRAMES project that defined the ETSI proposal for the UMTS standardization. From 1999 to 2003 he was with KPN Research, Leidschendam, The Netherlands where he was working on the radio optimization and performance estimation of the GPRS, UMTS and WLAN networks. At the same time he was affiliated with the Centre for TeleInfrastruktuur (CTiF) at Aaalbog University, Aalborg, Denmark where he has received his Ph.D. degree in 2008. Ljupcˇo is actively following 3GPP standardisation and has interactions with international fora such as e.g. NGMN. Ljupcˇo was involved in European IST research projects FP5 MOMENTUM (2001– 2004) on enhanced methods for UMTS planning, FP6 Ambient Networks (2004– 2007) on radio access selection in multi-radio access systems, and FP7 SOCRATES (2008–2010) on self-optimisation and self-healing for LTE wireless access networks. Dr. Jorguseski has one journal publications (IEEE Comm. Magazine), over 15 scientific publications in major conferences such as for example IEEE VTC-Fall, IEEE PIMRC, IEEE WPMC, etc. has contributed with editorship in two books on wireless tecommunications, and has co-authored two patents (and five pending patent applications). Pedro Sebastia˜o (S’95-M’05) received the B.Sc. degree in Electronics, Telecommunications and Computing from ISEL, Polytechnic Institute of Lisbon, Portugal, in 1992. He received his Licenciado and M.Sc. degrees in Electrical and Computer Engineering from IST, Technical University of Lisbon, in 1995 and 1998, respectively. He is a lecturer in LUI-ISCTE and research assistant at Instituto de Telecomunicac¸o˜es. He has been involved on several research projects and his interests include planning tools for Wi-Fi and WiMAX, stochastic models and efficient simulation algorithms for physical layer. Eng. Sebastia˜o is a member of IEEE, Sociedade Brasileira das Telecomunicac¸o˜es and Ordem dos Engenheiros. Rui Costa was born in Santare´m, Portugal, and received the Licenciado degree in Informatics from Instituto Matema´ticas e Gesta˜o - Universidade Lusofona in 1999. He is the responsible by the development and management of the Computer

About Authors

xv

Services and network of the Health Sciences Faculty of University of Beira Interior, he works on the development of wireless LAN, MAN and WAN, e-Learning, b-Learning and m-Learning applications, advanced WIA (wireless Internet applications), RIA (rapid Internet applications) in University environments for Web2. He is concluding is M.Sc. on cellular planning and deployment issues for fixed and portable WiMAX. Daniel Robalo received the Licenciado and M.Sc. degrees in Electrical Engineering from University of Beira Interior (UBI), Covilha˜, Portugal, in 2005 and 2008, respectively. His M.Sc. research was done in the context of the MobileMAN project, an internal project on WiMAX deployment from Instituto de Telecomunicac¸o˜es (IT). He is currently a researcher at IT, and his research interest is in the field of Broadband Wireless Access, including design and implementation of WiMAX networks. Besides conference papers, including IEEE ones, Eng. Robalo has two journal papers and one book chapter accepted for publication. Cla´udio Comissa´rio received the Licenciado and M.Sc. degrees in Electrical Engineering from University of Beira Interior (UBI), Covilha˜, Portugal, in 2007 and 2008, respectively. His M.Sc. research was done in the context of the MobileMAN project, an internal project on WiMAX deployment from Instituto de Telecomunicac¸o˜es (IT). Eng. Comissa´rio is currently with CELFINET, a Portuguese consultancy company for mobile operators. Anto´nio Rodrigues received the B.S. and M.S degrees in electrical and computer engineering from the Instituto Superior Te´cnico (IST), Technical University of Lisbon, Lisbon, Portugal, in 1985 and 1989, respectively, and the Ph.D. degree from the Catholic University of Louvain, Louvain-la-Neuve, Belgium, in 1997. Since 1985, Prof. Rodrigues has been with the Department of Electrical and Computer Engineering, IST, where is currently an Assistant Professor. His research interests include mobile and satellite communications, wireless networks, spread spectrum systems, modulation and coding. Hamid Aghvami is the Director of the Centre for Telecommunications Research at King’s College London. He is Managing Director of Wireless Multimedia Communications Ltd., his own consultancy company. Professor Aghvami leads an active research team working on numerous mobile and personal communications projects for third and fourth generation systems. He is a Fellow of the Royal Academy of Engineering, Fellow of the IET, Fellow of the IEEE, and in 2009 was awarded a Fellowship of the Wireless World Research Forum. Prof. Aghvami was a member of the Board of Governors of the IEEE Communications Society in 2001–2003, and has been member, Chairman, and Vice-Chairman of the technical programme and organising committees of a large number of international conferences. He is also founder of the International Symposium on Personal Indoor and Mobile Radio Communications (PIMRC). Oliver Holland obtained his B.Sc. degree (with First Class Honours) in Physics and Music from Cardiff University, and his Ph.D. in Telecommunications from King’s College London. Dr. Holland has jointly authored over 60 publications on a variety of topics, including several book chapters and one patent, and has served as an Organiser and Programme Committee member for a number of major

xvi

About Authors

conferences and journals. His interests include Beyond 3G Communications Systems, Spectrum Sharing, and Green Communications, among many others. Muhamad Kashif Nazir concluded his M.Sc. in Space Science at the University of Punjab Lahore in 2002. He is currently a “Signal Processing for Communications” M.Sc. student at King’s College London. His research is on “Modelling and Simulation of Efficient Broadband Wireless Access Architectures with Multi-hop Relays”, which also involved cost & revenue analysis for Fixed WiMAX.

Acknowledgements

First of all, the Editors would like to acknowledge Kirti Pasari from CTIF, Aalborg University for her big effort towards the completion of this book. Special thanks go to Mandalika Venkata Ramkumar and Dua Idris from CTIF for their hard effort in finalizing the manuscript. Authors also gratefully acknowledge Dr. Nicola Marchetti and Ambuj Kumar contributions and suggestions. The Editors acknowledge Thikrait Al Mosawi for the discussions and the Figure on Media independent Handover and IEEE 802.21. They also acknowledge B. Muquet, E. Biglieri, H. Sari and S. Ahamdi for using the following material from their papers: Figures 11.15 and 11.16, extracted from reference [18]/Chapter 11, and Table 12.11, extracted from reference [31]/Chapter 12. They also acknowledge IEEE for using the information from IEEE standards. Part of the work from Chapter 6 was conducted within the framework of the IST Sixth Framework Programme Integrated Project WEIRD (IST-034622), which was partially funded by the Commission of the European Union. Study sponsors had no role in study design, data collection and analysis, interpretation, or writing the book chapter. The views expressed do not necessarily represent the views of the authors’ employers, the WEIRD project, or the Commission of the European Union. We thank our colleagues from all partners in WEIRD project for fruitful discussions. The work from Chapters 8, 9 and 10 was partially funded by MobileMAN (Mobile IP for Broadband Wireless Metropolitan Area Network), an internal project from Instituto de Telecomunicac¸o˜es/Laborato´rio Associado, by CROSSNET (Portuguese Foundation for Science and Technology POSC project with FEDER funding), by “Projecto de Re-equipamento Cientı´fico” REEQ/1201/EEI/ 2005 (a Portuguese Foundation for Science and Technology project), by UBIQUIMESH, by the Marie Curie Intra-European Fellowship OPTIMOBILE (Cross-layer Optimization for the Coexistence of Mobile and Wireless Networks Beyond 3G, FP7-PEOPLE-2007-21-IEF) and by the Marie Curie Reintegration Grant PLANOPTI (Planning and Optimization for the Coexistence of Mobile and Wireless Networks Towards

xvii

xviii

Acknowledgements

Long Term Evolution, FP7-PEOPLE-2009-RG). Authors also acknowledge the COST Action 2100 – Pervasive Mobile & Ambient Wireless Communications. The authors acknowledge the fruitful contributions on ArcGIS tools from Eng Jose´ Roma˜o, Eng Jose´ Riscado and Prof. Victor Cavaleiro from STIG-UBI, and to the final year project students Hugo Carneiro, Jorge Oliveira, Dany Santos and Rui Marcos.

Contents

1

The Evolution Towards WiMAX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ramjee Prasad and Fernando J. Velez 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 History of Wireless Transmission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Broadband Access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.1 The Path Towards Wireless Broadband Access . . . . . . . . . . . . . . . 1.3.2 Types of Broadband Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.3 DSL Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Evolution of Wireless Broadband . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.1 Narrowband Wireless Local Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.2 First Generation Broadband Wireless Systems . . . . . . . . . . . . . . . . 1.4.3 Second Generation Broadband Wireless Systems . . . . . . . . . . . . . 1.4.4 Emergence of Global Standard Based Wireless Broadband . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 Fixed Wireless Access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5.1 Point-to-Point Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5.2 Point-to-Multipoint Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5.3 NLoS Point-to-Multipoint Networks . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5.4 Mesh Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6 Broadband Wireless Access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6.1 Licensed Frequency Bands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6.2 License-Exempt Frequency Bands . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.7 Background on IEEE 802.16 and WIMAX . . . . . . . . . . . . . . . . . . . . . . . . . . 1.7.1 IEEE 802.16 Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.7.2 IEEE 802.16a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.7.3 IEEE 802.16-2004 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.7.4 IEEE 802.16e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.8 Fixed and Mobile WIMAX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.9 Personal Broadband . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.10 WIMAX Forum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 1 3 11 11 13 15 17 17 17 18 19 20 20 21 22 22 23 23 24 25 26 26 27 27 30 32 34

xix

xx

Contents

1.10.1 Difference Between WiMAX Forum and IEEE 802.16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.10.2 Members . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.10.3 Working Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.10.4 WiMAX Products Certification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.10.5 WiMAX Forum Certified Program Milestones . . . . . . . . . . . . . 1.10.6 Certification Roadmap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.10.7 Benefits of an Industry Standard and Certified Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.10.8 WiMAX Forum Certified Products . . . . . . . . . . . . . . . . . . . . . . . . . 1.11 Frequency Band Allocations for WiMAX . . . . . . . . . . . . . . . . . . . . . . . . . . 1.12 ITU Classification Adds Credibility to WiMAX . . . . . . . . . . . . . . . . . . . 1.13 WiMAX Applications and Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.13.1 Fixed Broadband Wireless Applications . . . . . . . . . . . . . . . . . . . . 1.13.2 VoIP over WiMAX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.14 Salient Features of WiMAX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.15 Fixed, Portable, and Mobile Terminals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.16 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

OFDMA WiMAX Physical Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ramjee Prasad and Fernando J. Velez 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 History and Development of OFDM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2 Applications of OFDM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 OFDM Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.1 OFDM Versus FDM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.2 OFDM Signal Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 OFDM Transmission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.1 Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.2 Serial to Parallel Converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.3 Demodulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 OFDM Symbol Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6 ISI and ICI Mitigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7 Spectral Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.8 Orthogonal Frequency Division Modulation Access . . . . . . . . . . . . . . . . . 2.8.1 Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.8.2 Subchannelization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.8.3 OFDMA Subchannelization: Its advantages to WiMAX . . . . . 2.9 Advantages of OFDM Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.10 Disadvantages of OFDM Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.10.1 Strict Synchronization Requirement . . . . . . . . . . . . . . . . . . . . . . . . 2.10.2 Peak-to-Average Power Ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.10.3 Co-channel Interference Mitigation in Cellular OFDM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

34 34 35 35 41 41 42 44 45 48 48 49 50 55 57 60 61 63 63 64 64 67 67 67 69 71 71 73 73 74 74 77 77 77 77 78 78 80 80 80 81

Contents

xxi

2.11 Scalable OFDMA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 2.11.1 Parameters and Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 2.11.2 Reference Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 2.12 IEEE 802.16 PHY Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 2.12.1 Supported Frequency Bands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 2.12.2 Channel Bandwidth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 2.12.3 IEEE 802.16 PHY Interface Variants . . . . . . . . . . . . . . . . . . . . . . . 84 2.13 WirelessMAN-SC PHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 2.13.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 2.13.2 Duplexing Techniques and PHY Type Parameter Encodings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 2.13.3 Frame Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 2.13.4 Downlink PHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 2.13.5 Uplink PHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 2.14 WirelessMAN-OFDM PHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 2.14.1 Channel Coding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 2.14.2 Concatenated Reed–Solomon-Convolutional Code (RS-CC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 2.15 WirelessMAN-OFDMA PHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 2.15.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 2.15.2 Subcarrier Allocation Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 2.16 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 3

Medium Access Control Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ramjee Prasad and Fernando J. Velez 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 MAC Sub-layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 MAC Protocol Data Unit (PDU) and Service Data Unit . . . . . . . . . . . 3.4 Service Specific Convergence Sub-layer (CS) . . . . . . . . . . . . . . . . . . . . . 3.4.1 Mission and Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.2 Connections and Addressing, Connection ID (CID) and Service Flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 MAC Common Part Sub-layer (MAC CPS) . . . . . . . . . . . . . . . . . . . . . . . 3.5.1 Mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.2 MAC Addressing and Connections . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.3 MAC PDU Format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6 MAC Header Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6.1 Generic MAC Header . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6.2 MAC Header Without Payload . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7 MAC Sub-headers, Special Payloads and Sub-header Ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7.1 Significance of Type Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8 IEEE 802.16 MAC Connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

137 137 138 139 140 140 142 150 150 150 151 152 152 152 155 157 158

xxii

4

Contents

3.9 Basic, Primary and Secondary Management Connections . . . . . . . . . 3.9.1 MAC Management Message Format . . . . . . . . . . . . . . . . . . . . . . . . . 3.10 Fragmentation, Packing and Concatenation . . . . . . . . . . . . . . . . . . . . . . 3.10.1 Fragmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.10.2 Packing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.10.3 Concatenation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.10.4 CRC Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.11 Encryption of MAC PDUs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12 Automatic Repeat Request (ARQ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12.1 Error Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12.2 ARQ Blocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12.3 ARQ Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.12.4 HARQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.13 Scheduling Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.13.1 Scheduling Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.14 Bandwidth Allocation and Request Mechanisms . . . . . . . . . . . . . . . . . 3.14.1 Role of SSs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.14.2 Requests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.14.3 Grants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.14.4 Polling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.15 Contention Resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.15.1 Transmission Opportunity and Example for Contention-Based Bandwidth Requests . . . . . . . . . . . . . . . 3.16 Network Entry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.16.1 Downlink Channel Synchronization . . . . . . . . . . . . . . . . . . . . . . 3.16.2 Initial Ranging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.16.3 Capabilities Negotiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.16.4 Authentication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.16.5 Registration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.16.6 IP Connectivity Stages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.16.7 Transport Connection Creation . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.17 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

158 159 161 161 162 163 164 165 165 165 166 167 171 174 175 175 175 176 177 178 183

Quality of Service . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ramjee Prasad and Fernando J. Velez 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Interpretation and Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1 Technical and Customer-Specific QoS Parameters . . . . . . . . . . 4.2.2 QoS Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 QoS Architecture for IEEE 802.16 MAC Protocol . . . . . . . . . . . . . . . . 4.4 QoS Rovisioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.1 Service Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.2 Service Flow Attributes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

191

184 185 186 186 187 188 188 189 189 189 190

191 192 192 194 196 198 198 199

Contents

5

xxiii

4.5 Object Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5.1 Attributes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5.2 Global Service Flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5.3 Conventions Used for Global Service Class Name . . . . . . . . . . 4.5.4 Global Service Flow Class Name Parameters . . . . . . . . . . . . . . . . 4.5.5 Service Flow Creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6 Service Classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7 Authorization Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7.1 Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7.2 Static Authorization Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7.3 Dynamic Authorization Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8 Service Flow Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8.1 Provisioned Service Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8.2 Admitted Service Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8.3 Active Service Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.9 Service Flow Creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.9.1 Three-Way Handshaking Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.9.2 Dynamic Service Flow Creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.9.3 Dynamic Service Flow Modification and Deletion . . . . . . . . . . 4.10 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

199 199 201 201 201 203 204 205 205 205 205 206 206 207 207 208 208 209 210 214 214

Security Sublayer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ramjee Prasad and Fernando J. Velez 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Security Architecture for IEEE 802.16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1 Architecture and Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.2 Secure Encapsulation of MPDUs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.3 Encryption Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.4 Data Encryption Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.5 Key Management Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Authentication Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.1 RSA Authentication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.2 PKM EAP Authentication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4 PKM Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5 PKM Protocol MAC Management Messages . . . . . . . . . . . . . . . . . . . . . . 5.5.1 PKM Version 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6 Key Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.1 Authorisation Key (AK) Management . . . . . . . . . . . . . . . . . . . . . . . 5.6.2 SS Key Usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.3 PKM Version 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7 Network Aspects of Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7.1 Radius . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7.2 Extensible Authentication Protocol (EAP) . . . . . . . . . . . . . . . . . . .

215 215 216 216 218 218 219 222 223 223 224 224 224 227 240 240 244 246 247 248 248

xxiv

Contents

5.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 6

7

Mobility Management Architecture for WiMAX Networks . . . . . . . . Susana Sargento, Pedro Neves, Ricardo Matos, Marı´lia Curado, Bruno Sousa, Kostas Pentikousis, and Giada Landi 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 WiMAX Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.1 The WiMAX Forum Network Reference Model . . . . . . . . . . . . . 6.2.2 The IEEE 802.16 Reference Model . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.3 Mobility Management in WiMAX . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.4 IEEE 802.21 Media Independent Handover Overview . . . . . . 6.3 QoS-Aware Mobility Management: A Use Case for the WEIRD System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.1 Integrated QoS and Mobility Management . . . . . . . . . . . . . . . . . . 6.3.2 QoS-Aware Inter-technology Handovers for WiMAX . . . . . . . 6.4 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.1 MIHF Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.2 Link Layer Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.3 NSIS: QoS and MIH Transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.4 MM and CSC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.5 QoS Signalling and Resource Reservation . . . . . . . . . . . . . . . . . . . 6.4.6 Mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5 Testbed Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5.1 Handover Process Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5.2 Mobility Manager Initialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5.3 Resource Reconfiguration During HO Preparation and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5.4 MIH Transport Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

251

Radio Resource Management in WiMAX Networks . . . . . . . . . . . . . . . . . Jorguseski Ljupco and Ramjee Prasad 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 General Radio Resource Management (RRM) Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3 Radio Resource Definition in WiMAX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4 Downlink RRM Algorithms for WiMAX Systems . . . . . . . . . . . . . . . . 7.4.1 Reuse-One . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4.2 Reuse-Three . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4.3 Soft Re-Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4.4 Maximum C/I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4.5 Proportional Fair (PF) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

289

253 254 255 256 259 261 263 265 269 272 272 273 274 275 276 277 279 281 282 282 285 286 286

289 290 293 294 296 296 297 299 300

Contents

7.5 Simulation Evaluation of Downlink Sub-Carrier Allocation Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.6 Semi-analytical Evaluation of Downlink Sub-carrier Allocation Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.6.1 Comparison of Semi-analytical and Simulation Results . . . . . 7.7 Conclusions and Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

9

Radio and Network Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fernando J. Velez, Pedro Sebastia˜o, Rui Costa, Daniel Robalo, Cla´udio Comissa´rio, Anto´nio Rodrigues, and A. Hamid Aghvami 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Propagation Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.1 SUI Versus Modified Friis Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.2 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3 Cellular Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.1 Limitations of a Simplified Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.2 Signal-to-Noise-Plus-Interference Ratio . . . . . . . . . . . . . . . . . . . . . 8.3.3 Interference-to-Noise Ratio and Reuse Pattern . . . . . . . . . . . . . . 8.3.4 CNIR and Supported Throughput . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3.5 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4 Fixed WiMAX Planning Platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.1 Framework and Scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.2 Wireless Planning Tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4.3 Results with Adaptive Modulation and Coding Schemes . . . . 8.4.4 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . System Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fernando J. Velez, M. Kashif Nazir, A. Hamid Aghvami, Oliver Holland, and Daniel Robalo 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 CNIR Versus Physical Throughput Without Relays . . . . . . . . . . . . . . . 9.3 CNIR Versus Physical Throughput with Relays . . . . . . . . . . . . . . . . . . . 9.3.1 Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3.2 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3.3 DL Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3.4 UL Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4 Supported Physical Throughput . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.1 Implicit Function Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.2 Without Relays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.3 DL with Relays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.4 UL with Relays and K = 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.5 Use of Sub-Channelisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xxv

300 305 310 312 313 315

316 319 319 321 325 325 326 328 334 343 344 344 344 351 359 362 363 365

366 367 370 370 372 374 377 378 378 381 387 388 389

xxvi

Contents

9.5 Throughput with Sectorization, Relays and K = 1 . . . . . . . . . . . . . . . . . 390 9.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393 10

11

Business Models and Cost/Revenue Optimization . . . . . . . . . . . . . . . . . . . Fernando J. Velez, M. Kashif Nazir, A. Hamid Aghvami, Oliver Holland, and Daniel Robalo 10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Business Model Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3 Broadband Communications Business . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.4 Developing a WiMAX Business Plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.4.1 Market Research: Gathering the Input Parameters for a WiMAX Business Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.4.2 Market Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5 Range of Services Offered . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5.1 Classification of Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5.2 Varying Terrain Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.6 Cost/Revenue Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.6.1 Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.6.2 Hypothesis Without Relays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.6.3 Hypothesis and Assumptions with Relays . . . . . . . . . . . . . . . . 10.6.4 Optimization and Profit Without Relays . . . . . . . . . . . . . . . . . . 10.6.5 Optimization and Profit with Relays . . . . . . . . . . . . . . . . . . . . . . 10.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Multiple Antenna Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ramjee Prasad and Fernando J. Velez 11.1 Introduction to MIMO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1.1 MIMO Wireless Transmission System . . . . . . . . . . . . . . . . . . . . 11.1.2 Closed Loop MIMO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Capacity of MIMO Channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3 Multi-antenna in OFDM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4 Space Diversity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5 Space-Time Coding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5.1 Alamouti Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5.2 Second-Order STC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.6 Spatial Multiplexing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.6.1 Vertical Encoding SM & Horizontal Encoding SM . . . . . . 11.7 Collaborative Spatial Multiplexing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.7.1 Collaborative SM for UL PUSC . . . . . . . . . . . . . . . . . . . . . . . . . . 11.8 Diversity Gain and Array Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.9 Comparison of MIMO Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.10 System Model for MIMO OFDM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

395

396 397 397 398 398 400 404 404 405 406 406 408 409 411 416 419 421 423 423 425 426 427 427 429 429 430 432 432 433 435 435 436 438 440

Contents

11.11 Support of MIMO Technology in the IEEE 802.16 Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.11.1 Two-Antenna Downlink STC Transmission . . . . . . . . . . . . 11.12 Beamforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.12.1 Switched-Beam Antenna Arrays . . . . . . . . . . . . . . . . . . . . . . . . 11.12.2 Adaptive Antenna Systems (AAS) in IEEE 802.16e-2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.13 Trade-Off Between Spatial Multiplexing and Spatial Diversity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.14 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

WiMAX and Wireless Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ramjee Prasad and Fernando J. Velez 12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2 Advantages of WIMAX Over DSL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.3 Wi-Fi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.4 WiBRo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.5 ETSI HiperACCESS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.6 ETSI HiperMAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.7 iBURST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.8 Flash- OFDM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.9 IEEE 802.16 and 3G . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.10 3G Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.10.1 3GPP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.10.2 3GPP2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.11 HSDPA and HSUPA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.12 IEEE 802.16m and IMT-ADVANCED . . . . . . . . . . . . . . . . . . . . . . . . . . 12.13 IEEE 802.20/Mobile FI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.14 IEEE 802.21 and Media Independent Handover . . . . . . . . . . . . . . . . 12.15 IEEE 802.22 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.16 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xxvii

441 442 444 444 446 447 448 449 451 451 452 454 457 458 458 459 460 460 461 461 462 464 467 467 469 471 473 473

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477

.

Chapter 1

The Evolution Towards WiMAX Ramjee Prasad and Fernando J. Velez

Abstract Convergence is a step towards the unpredictable future of wireless communications. Important increases are foreseen in supported bit rates and remarkable improvements in services, applications and wireless communication components. After presenting a brief history of wireless communications, a vision on nowadays wireless ecosystem is presented and the path towards broadband wireless access is explored. Details on IEEE 802.16 evolution are given and the reasons for the existence of the WiMAX Forum are explained. WiMAX service classes are described and salient features of WiMAX are highlighted.

1.1

Introduction

The rapid growth of wireless communication and its pervasive use in all walks of life are changing the way we communicate in all fundamental ways. It is one of the most vibrant areas in the communication field today. Wireless communication dates back to the end of the nineteenth century when Maxwell showed through his equations that the transmission of information can be achieved without the need for a wire [1]. Later, experimentations by Marconi and other scientists proved that long distances wireless transmission may be a reality. True Wireless communications have gained a momentum in the last decade of twentieth century with the success of second Generation (2G) of digital cellular mobile services. Worldwide successes of Global System for Mobile Communications (GSM), Interim Standard 95 (IS-95), Personal digital Cellular (PDC) and digital Advanced Mobile Phone System (IS-54/136) have enabled pervasive ways of life for the new information and communication technology era. Second R. Prasad (*) Center for TeleInFrastruktur (CTIF), Aalborg University, Niels Jernes Vej 12, DK–9220 Aalborg Øst, Denmark e-mail: [email protected]

R. Prasad and F.J. Velez, WiMAX Networks, DOI 10.1007/978-90-481-8752-2_1, # Springer ScienceþBusiness Media B.V. 2010

1

2

R. Prasad and F.J. Velez

5G

GIMCV

2020 IMT-A/4G High speed WLAN

WiBro 802.16e

2010 + LTE

5 GHz WLAN

WiMAX

PN & PN Federation IEEE 802.22

WPAN IEEE 802.20

3G 2.4 GHz WLAN

Bluetooth

2000

ZigBee

2G

1990 1G

1980 GIMCV: Global Information Multimedia Communication Village

Fig. 1.1 The progress tree for communication technology [5, 6]

Generation (2G), 2.5G, and Third Generation (3G) standards of mobile systems are being deployed everywhere worldwide, with different versions of Universal Mobile Telecommunications System (UMTS), while efforts are going on towards the development and standardization of Beyond 3G (B3G) systems, for example, High Speed Packet Access (HSPA), Wireless Local and Personal Area Networks (WLANs/WPANs) and ultimately towards Fourth Generation (4G) [2–4]. Figure 1.1 illustrates how the progress towards the next generation in communication technology, 4G, can be perceived as a tree, with many branches. But this is not the end of the tunnel; ever increasing user demands have drawn the industry to search for always best connected solutions to support data rates of the range of tens of Mbps in a context of interoperability and cognitive radio. It will lead to 5G as shown in Fig. 1.1. This chapter is organized as follows. Section 1.2 addresses the history of wireless communications and presents convergence as a step towards the unpredictable future. Section 1.3 presents the path towards wireless broadband access, the types of broadband technologies and gives details on Digital Subscriber Line (DSL) broadband technology. Section 1.4 describes the evolution of wireless broadband, from narrowband wireless local loop to different generations of broadband wireless systems, towards the emergence of a global standard based wireless broadband. Section 1.5 addresses fixed wireless access. Section 1.6 present broadband wireless accesses and distinguishes between licensed and licensed-exempt frequency bands. Section 1.7 gives the background on IEEE 802.16 and Worldwide Interoperability for Microwave Access (WiMAX), namely IEEE 802.16 evolution,

1 The Evolution Towards WiMAX

3

from IEEE 802.16 to 802.16a, 802.16b, 802.16-2004, 802.16e and 802.16m as well as their characteristics. Section 1.8 compares fixed and mobile WiMAX. Section 1.9 presents the convergence of personal broadband while Section 1.10 describes the WiMAX Forum perspective. Section 1.11 presents the frequency band allocation for WiMAX while Section 1.12 gives details about ITU classification for WiMAX. Section 1.13 addresses WiMAX services and applications. Section 1.14 presents the salient features of WiMAX. Section 1.15 presents fixed, portable and mobile WiMAX terminals. Finally, Section 1.16 presents the conclusions.

1.2

History of Wireless Transmission

The notion of transmitting information without the use of wires seemed to be a magic in the nineteenth century. But in the year 1896 Marchese Guglielmo Marconi made it possible for the first time by demonstrating the ability of radio waves instant communication [1]. Since then new wireless communication methods and technologies have been evolving in the last over 100 years. The exciting history of Wireless can be divided into four periods, as shown in Tables 1.1–1.4: Table 1.1 Pioneer Era 1600 Dr. William Gilbert detects electromagnetic activity in the human body and describes it as “electricity”. 1837 Samuel F.B. Morse invents the Morse telegraph and sends messages over wires by using Morse code. 1865 Scientists, inventors, and hobbyists begin performing experiments with wireless. 1860s James Clark Maxwell’s EM waves postulates. 1880s Proof of the existence of EM waves by Heinrich Rudolf Hertz. 1905 First transmission of wireless and first patent of wireless communications by Gugliemo Marconi. 1909 Gugliemo Marconi received the Nobel prize. 1912 Sinking of the Titanic highlights the importance of wireless communications on the seaways. In the next years marine radio is established.

Table 1.2 Pre-cellular Era 1921 Detroit police department conducts field tests with mobile radio. 1933 In the United States, four channels in the 30–40 MHz range. 1938 In the United States, ruled for regular services. 1940 Wireless communications is stimulated by World War II. 1948 First commercial fully automated mobile telephone system is deployed in Richmond, United States. 1950s Microwave telephone and communication links are developed. 1958 A-Netz was introduced in Germany. 1960s Introduction of trunked radio systems with automatic channel allocation capabilities in the United States. 1970s Commercial mobile telephone system operated in many countries (e.g. 100 million vehicles on US highways, B-Netz in (West-)Germany).

4

R. Prasad and F.J. Velez

Table 1.3 Cellular Era/ Broadband Era [1–3]

1980s 1990s 2000s

Table 1.4 Convergence/ Personalisation Era

l l l l

2010+ 2015+ 2020’

Deployment of analogue cellular systems: 1G. Digital cellular deployment and dual mode operation of digital systems: 2G. International Mobile Telecommunications 2000 (IMT-2000)/deployment with multimedia services: 3G WPAN: Bluetooth, UWB WLAN: Wi-Fi WMAN: WiMAX, WiBro

Fourth generation (4G) Mega-communications Fifth generation (5G)

Pioneer Era Pre-cellular Era Cellular/Broadband Era Convergence/Personalisation Era

The pioneer era was marked by fundamental research and development, with contributions from Oersted, Faraday, Maxwell, Helmoltz, Rudolf Hertz, Righi, Lavernock, among many others. The history of modern wireless communications started in 1896 when Marconi submitted his first patent, the first one ever in the field of wireless telegraphy, and 1901, also with Marconi, who demonstrated wireless telegraphy by sending and receiving Morse code, based on long-wave (>>1 km wavelength) radiation, using high-power transmitters [4]. It had happened at the 11 December 1901. With a transmission range larger than 3,000 km, radio waves connected wirelessly Europe and America. How was it possible if the two terminals were not in Line-of-Sight (LoS)? The answer is the reflection onto the ionosphere, whose existence was postulated by Heaviside and Kennely. In 1907, the first commercial trans-Atlantic wireless service was initiated, using huge ground stations and 30100 m antenna masts. World War I saw the rapid development of communications intelligence, intercept technology, cryptography, and other technologies that later became critical to the advent of modern wireless systems. Later, Marconi discovered shortwave ( 11 GHz. Feb 2002 Korea allocates spectrum at 2.3 GHz for wireless broadband (WiBro). Jan 2003 IEEE 802.16a standard completed. June 2004 IEEE 802.16-2004 standard completed and approved. Sept 2004 Intel begins shipping the first WiMAX chipset, called Rosedale. Dec 2005 IEEE 802.16e standard completed and approved. Jan 2006 First WiMAX Forum–certified product announced for fixed applications. June 2006 WiBro commercial services launched in Korea.

20

R. Prasad and F.J. Velez

1.5

Fixed Wireless Access

Three options are popular today for design of wireless networks – Point-to-Point (PtP), Point-to-Multipoint (PtM), and Mesh topologies. The types of fixed wireless network topologies can be categorized into four types: l l l l

PtP networks PtM networks NLoS PtM networks Mesh networks

1.5.1

Point-to-Point Networks

As the name implies, a point-to-point wireless network is a direct link between two distinct locations. In the diagram, PtP connections are represented by the red lines. These connections are commonly used in cellular backhaul (from the Base Station, BTS, site towards the network operations centre) and for building-to-building extensions of IP and circuit-switched services (i.e. analogue PBX). Fiber optics and leased copper connections are examples of “wired” PtP networks. PtP networks consist of one or more fixed PtP links, usually employing highly directional transmitting and receiving antennas, as illustrated in Fig. 1.10. Networks of such links connected end to end can span great distances. Links connected end to end are often referred to as tandem systems, and the analysis for the end-to-end reliability or availability of the whole network must be calculated separately from the availability of individual links [15].

Line of Sight (LOS) Non Line of Sight (NLOS)

Not optimal some loss of signal

Base Station

Fig. 1.10 PtP network connecting two cities through mountaintop repeaters

1 The Evolution Towards WiMAX

1.5.2

21

Point-to-Multipoint Networks

PtM networks used a ‘hub and spoke’ approach to deliver data services as illustrated in Fig. 1.11. The hub is analogous to the base station in a cellular system. It consists of one or more broad-beam antennas that are designed to radiate toward multiple end-user terminals. Depending on the frequency band employed, and the data rates to be provided to end users, normally several hubs are needed to achieve ubiquitous service to a city. The remote end-user terminals are engineered installations in which directional antennas have been installed in locations that are in the LoS to the hub and oriented by a technician to point at the hub location. In some cases this may require extensive work at each terminal location. PtM network architecture is by far the most popular approach to fixed broadband wireless construction. It mimics the network topology successfully used for decades in wired telephone networks, cable television networks, and even electrical, gas, and water utilities of all sorts. For wireless, the major drawback is the cost of the infrastructure to construct the hubs needed to achieve comprehensive LoS visibility to a large percentage of the service.

Subscriber Station (SS)

Fig. 1.11 PtM network

22

1.5.3

R. Prasad and F.J. Velez

NLoS Point-to-Multipoint Networks

NLoS PtM networks are identical in topology to the PtM networks described above. The difference lies in the nature of the remote terminals. Instead of the remote terminals being engineered and professionally installed to achieve successful performance using an outside antenna, the terminals are arbitrarily positioned at the convenience of the end user inside a house or office. In most cases, the location of these terminals will be places that do not have a clear, obstruction-free view of a network hub and are thus called NLoS. The signal attenuation and amplitude variability that occurs along the wireless signal path from the network hub to NLoS location present new challenges to system designers in their efforts to provide a reliable high-speed data service to every terminal.

1.5.4

Mesh Networks

Mesh networks are a relatively new, evolving type of wireless broadband technology that may enable more flexible and more efficient expansion of wireless broadband services, Fig. 1.12. In the mesh topology each node has redundant connections to other nodes in the network, as shown in the figure. Unlike traditional Wireless Metropolitan area Networks (WMANs) or Wireless Local Area networks (WLANs), in which each “node” (or consumer device) in the network communicates only with a central antenna or base station, in a mesh network, each node

Fig. 1.12 Mesh networks

1 The Evolution Towards WiMAX

23

can function as an access point and transmit information to other nodes in close proximity. If one node goes out of service, the other nodes will route the traffic around it, making mesh networks a relatively robust communications technology. One of the key aspects of a mesh network is the routing functionality of its nodes which allows them to take the best route in communication with other nodes or networks. When a backhaul-based network is deployed in mesh mode, it does not only increase the wireless coverage, but it also provides features such as lower backhaul deployment cost, rapid deployment, and reconfigurability. Various deployment scenarios include citywide wireless coverage, backhaul for connecting 3G RNC with base stations, and others.

1.6 1.6.1

Broadband Wireless Access Licensed Frequency Bands

International Telecommunications Union (ITU) regulates the use of radio spectrum worldwide, which operates with the participation of all member nations [16]. The spectrum available for the construction of broadband wireless systems can be divided into licensed and license-exempt frequency bands. In general, licensed spectrum provides for some degree of interference protection because each new licensee must demonstrate compliance with certain standards for limiting interference to other existing nearby licensed systems. There are also radiated transmitter power level and other parameter limitations that each licensee must observe. License-exempt bands do not require individual transmitters to be licensed in order to operate, but there are still radiated power restrictions that usually keep power at low levels as a de facto way of limiting interference. There may also be a rudimentary channelization scheme and modulation standard; again, to make possible as many successful operations as possible without destructive interference. Some cooperation and coordination may sometimes be necessary to make the most of these measures. Cordless telephones, remote control toys, and IEEE802.11b/802.11a wireless LAN devices are examples of license-exempt systems. There are a number of frequency bands that have been allocated throughout the world for use by licensed fixed broadband services. Within the general ITU band designations, individual countries may elect to implement or not implement policies that allow those frequencies to be licensed and used within their country boundaries. This is especially true for fixed broadband wireless services. However, Table 1.6 provides a convenient summary for most of the European countries. The frequency bands listed are intended as examples of the variety of services that have access to the microwave spectrum for fixed services. The table include the major bands used for newer PtP and PtM broadband services such as Local Multipoint Distribution Service (LMDS).

24

R. Prasad and F.J. Velez

Table 1.6 European licensed fixed wireless bands [16–19] Frequency band Description (GHz) 3.4–3.6 Duplex spacing’s of 50 or 100 MHz are employed. 3.7 GHz is the upper limit of this band is in some countries. 3.8–4.2 High capacity public operator band for PtP link systems. 5.9–7.1 High capacity public operator band for PtP link systems. 7.1–8.5 Medium and high capacity public operator band for long haul PtP systems. 10.15–10.65 5  30-MHz channels with duplex spacing of 350 MHz. 10.7–11.7 High capacity public operator band for PtP link systems. 12.7–13.3 Low and medium capacity public operator band. 14.4–15.4 Fixed link operations of various types. 17.7–19.7 Public operator band for low and medium capacities. 21.2–23.6 Public operator band for PTP link systems of various types. 24.5–26.5 ETSI 26-GHz band. 3.5- to 112-MHz FDD channels with 1008-MHz duplex spacing. 37–39.5 Common carrier band for PtP link systems.

In addition to requirements to obtain a license for systems operating in these bands, each band also has a number of technical criteria that each system must satisfy. In general, these criteria are established to reduce or minimize interference among systems that share the same spectrum, and to ensure that the spectrum efficiency is sufficiently high to justify occupying the spectrum. In a given band, there may be requirements for minimum and maximum radiated power levels, particular efficient modulation types, and even standards for the radiation patterns of directional antennas to reduce interference to other operators in the band. These technical standards can be detailed and complex, and may vary from country to country. Designing and deploying a fixed wireless system in any particular country requires a careful review and functional understanding of the administrative rules that govern the use of the intended licensed spectrum space.

1.6.2

License-Exempt Frequency Bands

There is a growing interest in using the so-called license-exempt bands. One of the primary reasons is that it allows users of the wireless service to purchase offthe-shelf wireless modems for connecting to a system. In the United States, the 11 Mbps IEEE 802.11b standard that specifies Direct Sequence Spread Spectrum (DSSS) technology operating in the 2.4 GHz band is the best current example of self deployed license-exempt technology. However, license-exempt bands offer no regulatory interference protection except that afforded by the interference immunity designed into the technology itself. With relatively modest penetration of these systems to date, the robustness of the design for providing the expected quality of service in the presence of widespread interference and many

1 The Evolution Towards WiMAX

25

Table 1.7 US license-exempt fixed wireless bands [16–19] Frequency band Service Description (GHz) name 2.4–2.483 ISM Industrial, Scientific and Medical (ISM). This is the band for IEEE 802.11b DSSS networks operation 5.15–5.35 U-NII Unlicensed National Information Infrastructure (U-NII). This band is where IEEE 802.11a OFDM systems operate among several other proprietary standards. Channel bandwidth is 20 MHz. Particular power limits apply for segments of this band, intended for indoor and outdoor applications of are 20 MHz 5.725–5.825 GHz U-NII Same as 5.15–5.35-GHz U-NII band except that this band is intended only for outdoor applications with radiated power levels up to 4 W

Table 1.8 European license-exempt fixed wireless bands [16–19] Frequency band Service name Description (GHz) 2.4–2.483 Wi-Fi Industrial, Scientific and Medical (ISM). This is the same band where IEEE 802.11b DSSS networks operate in the United States 5.15–5.35 HiperLAN HiperLAN is the fast wireless network standard for Europe, which uses an OFDM transmission standard similar to IEEE 802.11a. This band is intended for indoor operations with radiated powers limited to 200 mW 5.470–5.725 GHz HiperLAN/2 Proposed frequency band for outdoor operation with radiated power levels limited to 1 W

contending users has yet to be fully tested. As the number of people using licenseexempt equipment increases in a given area, the ultimate viability of having a multitude of people using a limited set of frequencies will be tested. Table 1.7 shows the license-exempt bands currently used in the US for fixed broadband communications. The license-exempt spectrum has been designated in Europe, though the uptake of the technology has been slower than in the United States. As discussed in the next section, several long-running standard-setting efforts designed for this purpose did not bear fruit in a timely fashion, resulting in many of these efforts being suspended or abandoned in favour to US standards already in place. Table 1.8 shows the license-exempt bands currently available for use in Europe.

1.7

Background on IEEE 802.16 and WIMAX

The 802.16 activities were initiated in August 1998 in a meeting called by the National Wireless Electronics Systems Testbed (N-WEST) of the U.S [20] National Institute of Standards and Technology. The effort was welcomed in IEEE 802,

26

R. Prasad and F.J. Velez

which opened a Study Group. The 802.16 Working Group has held weeklong meetings at least bimonthly since July 1999.

1.7.1

IEEE 802.16 Evolution

The IEEE 802.16 group was formed in 1998 to develop an air-interface standard for Broadband Wireless Access (BWA) and to support the development and deployment of wireless metropolitan area networks, Table 1.9. The main task of the Working Group is addressing primarily applications of wireless technology to link commercial and residential buildings to high-rate core networks and thereby provide access to those networks. This link is known popularly as the “last mile”. The evolution of the IEEE 802.16 standard is shown in Table 1.9. The standard has adapted many of the concepts from the popular cable modem Data Over Cable Service Interface Specification (DOCSIS) standard related to the MAC layer. The group released the first standard in December 2001. This standard was designed for 10 GHz–66 GHz fixed frequencies that require LoS conditions as in LMDS.

1.7.2

IEEE 802.16a

The IEEE 802.16 group subsequently produced 802.16a, an amendment to the standard, to include NLOS applications in the 2–11 GHz band instead of LOS 266 GHz. The significant difference between these two frequency bands lies in the ability to support Non-Line-of-Sight (NLOS) operation in the lower frequencies, something that is not possible in higher frequency bands [20, 21]. Consequently, the 802.16a amendment to the standard opened up the opportunity for major changes to the PHY layer specifications specifically to address the needs of the 2–11 GHz bands. This is achieved through the introduction of three new PHY-layer specifications (a new Single Carrier PHY, a 256 point FFT OFDM PHY, and a 2048 point FFT OFDMA PHY); major changes to the PHY layer specification as compared to Table 1.9 The evolution of the IEEE 802.16 standard [20] Date IEEE standard Description Dec 2001 802.16 First Standardisation Using 10–66 GHz spectrum band, only LoS Only Point to Multipoint configuration supported Jan 2003 802.16a Amendment that supports NLoS using 2-11 GHz band Both PtM and mesh configuration supported OFDM(A) are adopted as principal PHY implementations Oct 2003 802.16b QoS Provisioning Oct 2004 802.16-2004 This revised standard replaced previous versions Dec 2005 802.16e Enhancements for mobility support (Mobile WiMAX)

1 The Evolution Towards WiMAX

27

the upper frequency, as well as significant MAC-layer enhancements. Additions to the MAC layer, such as support for Orthogonal Frequency Division Multiple Access (OFDMA), were also included.

1.7.3

IEEE 802.16-2004

This standard was the first practical standard of the IEEE 802.16 family. It is popularly called as fixed WiMAX, the name of the industrial alliance for its production. IEEE 802.16-2004 air interface standard was published in October 2004. This version replaced all previous versions of the IEEE 802.16 standard including 802.16-2001, 802.16a-2003, 802.16c and 802.16-REVd, Table 1.10. It integrated all the previous standards and re-edited PHY and MAC layer contents to improve the system performance. Although IEEE 802.16-2004 specifies a standard for fixed access, the actual applications supports nomadicity. Nomadicity means that user devices can move as long as they do not operate while doing so. The standard is specified to allow nomadicity, where users can access the service from various locations covered by the network. The complete edition of IEEE 802.16-2004 was approved in December 2004. The types of PHY layers and the respective frequency bands in which they are operating are shown in Table 1.10.

1.7.4

IEEE 802.16e

In December 2005, the IEEE group completed and approved IEEE 802.16e-2005, an amendment to the IEEE 802.16-2004 standard that added mobility support. The IEEE 802.16e-2005 forms the basis for the WiMAX solution for nomadic and mobile applications and is often referred to as mobile WiMAX [23]. Mobility has become the necessity of time and will create a new market for mobile broadband services. With the introduction of 802.16e standard seamless handover and roaming are possible when users move from one cell site area to another. It adds 128, 512 Table 1.10 IEEE 802.16-2004 nomenclature [20, 22] Frequency spectrum Physical layer type 10–66 GHz WirelessMAN-SC 2*802.16e Up to 350 km/h

802.16i — Mobile Management Information Base (Mobile MIB) 802.16j — Multihop Relay Specification 802.16Rev2 — Consolidate 802.16-2004, 802.16e, 802.16f, 802.16g and possibly 802.16i into a new document

There will also be a future development towards the air interface, the IEEE 802.16m. The data rates will be 100 Mbps for mobile applications and 1 Gbps for fixed applications, cellular, macro and micro cell coverage. Nowadays, there is no restrictions on the RF bandwidth (which is expected to be 20 MHz or higher) [28]. Table 1.12 shows the IEEE 802.16m requirements.

30

1.8

R. Prasad and F.J. Velez

Fixed and Mobile WIMAX

Mobile WiMAX, referred to by the standard 802.16e, adds mobility to the WiMAX specifications. PC data cards, mobile handsets and laptops with embedded WiMAX chips are being planned by vendors on this standard. In countries where regulation prohibits full mobility for alternative wireless technologies such as WiMAX, operators can also deploy 802.16e networks for fixed and nomadic access. The Telecom industries faith in this new technology is reflected in the decision of leading chip vendors to invest heavily into the manufacturing of miniaturized 802.16e chipsets. Economies of scale are expected to drive price once large scale production starts. A comparison between fixed and mobile WiMAX is shown in Table 1.13 while Fig. 1.13 illustrates the concept. HOW DOES IT WORK? WiMAX uses microwave radio technology to connect computers to the Internet in place of wired connections such as DSL or cable modems, Fig. 1.14. WiMAX works very much like cell phone technology in that reasonable proximity to a base station is required to establish a data link to the Internet. Users within 3–5 miles of the base station will be able to establish a link using NLoS technology with data rates as high as 75 Mbps. Users up to 30 miles away from the base station with an antenna mounted for LoS to the base station will be able to connect at data rates approaching 280 Mbps. IEEE 802.16-based broadband and mobile wireless access is expected to be a significant component in the next-generation wireless systems. The IEEE 802.16 standard, which incorporates several advanced radio transmission technologies such as orthogonal frequency-division multiplexing (OFDM), adaptive modulation and coding (AMC), and adaptive forward error correction (FEC), is designed to

Table 1.13 Fixed and mobile WiMAX comparison [20, 24, 25] Parameter Fixed WiMAX Mobile WiMAX Definition Allows BWA when the user is in the Allows for handover of a call/data range of a WiMAX BS but not session as the user moves when the user is roaming. between radio towers Standard 802.16d 802.16e Release date Q3 2004 December 2005 Frequency 2–66 GHz Licensed bands 2–6 GHz l Allows for line of sight as well l Amendment for mobile wireless Key feature as non-line of sigh deployments broadband up to vehicular l Selectable channel bandwidth speeds l At vehicles, user data rates will be ranging from 1.25 to 20 MHz lower than for pedestrians Application Fixed and Nomadic Access Supports mobile and Nomadic Wireless Access CPE requirement Outdoor directional antenna, indoor PC data cards, laptops and mobile modems handsets with embedded CPE Required WiMAX chips

1 The Evolution Towards WiMAX

31

Fixed WiMAX

Mobile WiMAX

Embedded WiMAX Laptopts

Wi-Fi

Indoor Modem

Outdoor CPE

WiMAX PC Cards WiMAX Mobile Phone

Fig. 1.13 Fixed and mobile WiMAX

Nomadic Broadband complementary to 3G, EDGE & Wi-Fi

Broadband Access for Enterprise

802.16-e

802.16-2004

Broadband Access for Public hotspots 802.16-2004 004

Broadband Access @ Home complementary to DSL & Cable 802.16-2004

Wi-Fi

Fig. 1.14 WiMAX operation

provide broadband wireless capability using a well defined quality of service (QoS) framework. Therefore, this is a promising technology to provide wireless services requiring high-rate transmission (in the range of tens of megabits per second) and strict QoS requirements in both indoor and outdoor environments.

32

R. Prasad and F.J. Velez

Similar to the Wi-Fi Forum for the IEEE 802.11 WLAN standard, the WiMAX (worldwide interoperability for microwave access) Forum, which is a non-profit organization, encourages and supports IEEE 802.16-based broadband wireless access (BWA). The main role of the WiMAX Forum is to standardize and maintain the process of testing and the certification program for compatibility and interoperability of IEEE 802.16 equipment. IEEE 802.16/WiMAX technology intends to provide broadband connectivity to both fixed and mobile users in a wireless metropolitan area network (WMAN) environment. To provide flexibility for different applications, the standard supports two major deployment scenarios: Last-mile BWA: In this scenario, broadband wireless connectivity is provided to home and business users in a WMAN environment. The operation is based on a point-to-multipoint single- hop transmission between a single base station (BS) and multiple subscriber stations (SSs). Backhaul networks: This is a multihop (or mesh) scenario where a WiMAX network works as a backhaul for cellular networks to transport data/voice traffic from the cellular edge to the core network (Internet) through meshing among IEEE 802.16/WiMAX SSs.

1.9

Personal Broadband

Personal broadband expands the availability of a true broadband connection beyond the home and the office, allowing subscribers to have access to the same level of service and the same applications regardless of location. Figure 1.15 presents the evolution of broadband connectivity around the world in residential and commercial settings. 4.5 4 3.5 MidEast/Africa

3

Asia

2.5 2

Europe

1.5

Americas

1 0.5 0 2003

2004

2005

2006

2007

2008

Fig. 1.15 Accelerating demand for broadband connectivity around the world in residential and commercial settings continues to grow (From Intel Capital)

1 The Evolution Towards WiMAX

33

Fixed WiMAX based on IEEE 802.16-2005

Mobile WiMAX based on IEEE 802.16-2004 Fixed Access Portable Access Mobile Access

Fig. 1.16 Personal broadband requires a wireless interface that supports fixed, portable & mobile access

Depending on the technologies deployed, personal broadband may provide less throughput than its wired broadband last mile technologies, or not be as ubiquitous as cellular networks. Personal broadband, however, brings a precious advantage: convenience. With a single subscription, personal broadband subscribers can get online regardless of location, without having to worry about throughput, service availability, or metered charges at some locations. Personal broadband is located at intersection of the evolution trends in Internet access, from dial-up to always-on broadband to Wi-Fi wireless network, and the shift towards mobility in voice communications, which in the US now account for more minutes than the fixed lines for the average subscriber: subscribers value a flexible service that follows them wherever they go. Some 3G users with flat-fee plans have started to use their subscription as a sort of personal broadband service: they keep their 3G connection active regardless of location and, when they need more bandwidth and they are at home or in the office, they switch to the wired network, Fig. 1.16. In the current 3G networks, this is only possible because there are still very few users using the service and therefore congestion is not a problem. At the current price levels of $60–80 per month and above, this is also clearly a service aimed at a small niche of business users that are relatively price insensitive. Personal broadband is aimed at a wider subscriber base that includes both business and consumer users and therefore needs to rely on technologies that are scalable and costeffective, and provide high throughput. It is still too early to know which applications personal broadband subscribers will use, but one thing is clear: there will be no killer application. Instead, the appeal of personal broadband is to make available all the existing applications everywhere. Web surfing, VoIP, email and VPN connectivity, downloads are likely to be among the most popular applications. Service providers do not need to develop specific applications or content before rolling out the service, as they did for cellular data services, where the throughput limitations have required the development of optimized applications. This removes a big burden from service providers which have been to date not very successful at this task and allows them to focus on their core business.

34

1.10

R. Prasad and F.J. Velez

WIMAX Forum

1.10.1 Difference Between WiMAX Forum and IEEE 802.16 Wireless Interoperability for Microwave Access (WiMAX) is an industrial forum that is responsible for the task of ensuring interoperability and conformance among the systems and solutions developed by various vendors based, on IEEE 802.16 standard for Broadband Wireless Access (BWA). WiMAX Forum is coalition of vendors, service providers, technology providers, and manufacturers, etc. Its union symbolises industry leaders who are committed to promote broadband wireless access based on WiMAX technology. The WiMAX forum is a non profit organization and was formed in July 2001. This forum defines and conducts interoperability tests and awards the equipment manufacturers with the ‘WiMAX Certified’ label [20]. WiMAX Forum and WiMAX Forum Certified are both trademark names of an industry trade organisation. The difference between WiMAX and IEEE 802.16 is that the WiMAX Forum is a non-profit trade association industry group. The mission of IEEE 802.16, ETSI HIPERMAN and the WiBro working groups is to formulate the specifications. In contrast the aim of the WiMAX Forum is to incorporate the variations in the specifications of the three different working groups to ensure interoperability amongst them. This ultimately would benefit the end users who can buy the brand of their choice, with all the features they are interested in and with the full confidence that this would work with all other certified products. The IEEE Working Group 802.16 is responsible for the development of the 802.16 standard including the air interface for Broadband Wireless Access. The activities of this working group were initiated in a meeting in August 1998, called by National Wireless Electronics Systems Testbed (N-WEST) which is a part of U.S. National Institute of Standards and Technology [20]. Initially the group focused on the development of standards and air interface for the 10–66 GHz band. Later an amendment project led to the approval of the IEEE 802.16a standard meant for 2–11 GHz band. The final approval of the 802.16a Air Interface specification came in January 2003 [30]. It comprises of Industry leaders who are committed to promote broadband wireless access based on WiMAX technology.

1.10.2 Members The WiMAX Forum has more than 522 members [30] comprising the majority of operators, component and equipment companies in the communications ecosystem, Fig. 1.17.

1 The Evolution Towards WiMAX

35

Content / Ecosystem

Component Suppliers

WiMAX Forum Members

Service Providers

System Vendors

Fig. 1.17 Composition of WiMAX forum members [30]

1.10.3 Working Groups The WiMAX Forum has organized the Working Groups shown in Table 1.14 to address critical areas of focus in bringing WiMAX Forum Certified products to the marketplace. The WiMAX Forum works closely with service providers and regulators to ensure that WiMAX Forum Certified systems meet customer and government requirements. “WiMAX Forum” is a registered trademark of the WiMAX Forum. “WiMAX”, the WiMAX Forum logo, “WiMAX Forum Certified”, and the WiMAX Forum Certified logo are trademarks of the WiMAX Forum. All other trademarks are the properties of their respective owners [30].

1.10.4 WiMAX Products Certification WiMAX Forum is the exclusive organization dedicated to certifying the interoperability of products based upon IEEE 802.16/ETSI HiperMAN. It defines and conducts conformance and interoperability testing to ensure that different vendor systems work seamlessly with one another. Those that pass conformance and interoperability testing will receive the “WiMAX Forum Certified™” designation [26, 30]. Today in the market there are many vendors claiming their equipment is “WiMAX-like”, WiMAX-compliant”, etc. These products do not have WiMAX Forum Certified on it, which means that their equipment is not independently

36

R. Prasad and F.J. Velez

Table 1.14 Working groups of WiMAX forum Working group name Scope Application Working Define applications over WiMAX that are necessary to meet core Group (AWG) competitive offerings and that are uniquely enhanced by WiMAX Certification Working Handles the operational aspects of the WiMAX Forum Certified Group (CWG) program Evolutionary Technical Maintains existing OFDM profiles, develops additional fixed OFDM Working Group profiles, and develops technical specifications for the evolution (ETWG) of the WiMAX Forum’s OFDM based networks from fixed to nomadic to portable, to mobile Global Roaming Assures the availability of global roaming service for WiMAX Working Group networks in a timely manner as demanded by the marketplace (GRWG) Marketing Working Promotes the WiMAX Forum, its brands and the standards which form Group (MWG) the basis for worldwide interoperability of BWA systems Network Working Creates higher level networking specifications for fixed, nomadic, Group (NWG) portable and mobile WiMAX systems, beyond what is defined in the scope of 802.16 Regulatory Working Influences worldwide regulatory agencies to promote WiMAXGroup (RWG) friendly, globally harmonized spectrum allocations. Service Provider Gives service providers a platform for influencing BWA product and Working Group spectrum requirements to ensure that their individual market needs (SPWG) are fulfilled Technical Working The main goal of the TWG is to develop technical product Group (TWG) specifications and certification test suites for the air interface based on the OFDMA PHY, complementary to the IEEE 802.16 standards, primarily for the purpose of interoperability and certification of Mobile Stations, Subscriber Stations and Base Stations conforming to the IEEE 802.16 standards

certified to be interoperable with other vendors’ equipment. Only WiMAX Forum Certified™ equipment is proven interoperable with other vendors’ equipment. The logo “WiMAX Forum Certified” means users can buy a product or service based on the IEEE 802.16 standard from different companies and be confident that everything will work together. This approach is similar to what 802.11-based Wi-Fi networks used and has proved to be very successful. The IEEE 802.11 standard set the requirements, and the Wi-Fi Alliance ensured product compliance for interoperability. The WiMAX Forum manages the certification process by developing and adopting the relevant specifications, establishing the process and test suites to be used, selecting the testing labs, issuing the certification certificates, and maintaining a registry of WiMAX Forum Certified products. However, it is not directly involved in certification testing; this keeps the program independent and gives vendors the freedom and flexibility to choose their preferred lab. Vendors primarily work with the WFDCL of their choice, which accepts the certification application, conducts the tests, and, if tests are successful, issues a certification certificate. Current certification labs are located in China, Spain, South Korea, Taiwan, and the USA listed in Table. The WiMAX Forum is evaluating additional labs in Brazil, Japan, and India.

1 The Evolution Towards WiMAX

1.10.4.1

37

Certification Process

The WiMAX Certification label, guarantees both WiMAX/802.16-compliance and equipment interoperability. The certification process is divided into the development of conformance testing and interoperability testing. Development of the certification program is done by the WiMAX Certification Working Group (CWG). Certification testing is intended only for complete systems such as a BS or an SS, not individual solution components such as radio chips or software stacks. The introduction of BS/SS reference designs may also be considered for testing to show that the design conforms to the IEEE 802.16 specification and is interoperable with other WiMAX Forum Certified equipment. Conformance testing is a process where BS and SS manufacture test units to ensure that they perform in accordance with the specifications. Conformance test ensures interoperability with other WiMAX equipment and a positive end-user experience for your customers. WiMAX conformance testing can be done by either the certification lab or another test lab. At the lab, the manufacturers (both BS and SS) test their pre-production or production units to ensure that they perform in accordance with the specifications called out in the WiMAX Protocol Implementation Conformance Specification (PICS) document. Based on these results manufacturers may choose to modify their hardware and/or firmware and formally re-submit these units for conformance testing. WiMAX interoperability involves multi-vendor test process .This is hosted by the certification lab. Interoperability testing to verify that equipment from different vendors works within the same network. At least three vendors have to submit equipment within the same certification profile (defined by spectrum band, channelization and duplexing) to start interoperability testing.

1.10.4.2

Certification Testing

WiMAX Forum conducts Certification testing in independent laboratories. Table 1.15 lists the various certification laboratories termed as WiMAX Forum Designated Certification Laboratories (WMDCLs). The certification methodology is defined by the WiMAX Forum while the testing is conducted at the lab chosen by the vendor. Upon successful completion Table 1.15 WiMAX Forum Designated Certification Laboratories (WMDCLs)

AT4 Wireless (Spain, USA) www.at4wireless.com Bureau Veritas ADT (Taiwan) www.adt.com.tw CCS/TTC (Taiwan) Compliance Certification Services (CCS) www.ttc.org.tw China Academy of Telecommunication Research (China) www.catr.cn Telecommunications Technology Association (South Korea) www.tta.or.kr

38

R. Prasad and F.J. Velez

of all tests, vendors receive a WiMAX Forum Certified certificate and a test report, and can list their certified equipment on the WiMAX Forum Certified Product Registry. Certification testing is performed against the backdrop of the following test suites/purposes: l

l

l

Protocol Implementation Conformance Statement (PICS) Proforma — Is a document listing the required, optional and conditionally required features of a device assigned to conform to a standard. This document helps determine whether a device intends to meet all or a subset of the requirements for conformance with the view of helping to direct the test lab as to which tests to perform. Test Suite Structure & Test Purposes (TSS&TP) — All the tests that should be performed for all system options are listed in this document. It is structured in such a way that testing like topics are grouped together. The test “purposes” include the initial condition, stimulus and expected results but not necessarily the detailed steps. In addition to covering the requirements, it should be structured such that answers given by the manufacturer in the PICS Proforma can be mapped to the set of tests that need to be run for a particular device. Abstract Test Suite (ATS) — This consists of the detailed test procedures for the test purposes outlined in the TSS&TP. The ATS is abstract in that it’s usually written in a test language, such as TTCN8, for portability amongst a variety of test labs rather than containing specific detailed instructions for specific models of test equipment.

1.10.4.3

Plugfest

The Plugfest is a preview of full interoperability testing which allows vendors to get an early look at how well their equipment interoperates, Fig. 1.17. The plugfest conducted by WiMAX Forum generally last for 1 week at contracted testing site. Equipment is validated and verified “interoperable” when it is tested to be interoperable with other vendors’ equipment. This is done by demonstrating that their respective hardware is able to send and receive packets with two other vendors involving base stations and subscriber stations for a selected. Before the Plugfest, participating vendors must agree on a set of RF/PHY characteristics within a given certification profile. In all instances, a minimum of three vendors must be available to conduct interoperability testing in a certification profile. Ideally, the WiMAX Forum requires a minimum of five to six vendors to execute the planning of a Plugfest. WiMAX equipment can operate at any frequency below 11 GHz, using channel bandwidths ranging from 1.75 to 10 MHz, and with both TDD and FDD duplexing1. To make interoperability effective and testable, the WiMAX Forum has created system profiles and certification profiles (Fig. 1.18) that define classes of products. The WiMAX system profiles are as follows: l

System profiles are based on versions of the IEEE 802.16 and ETSI HiperMAN standards and define the key mandatory and optional features that are tested in

1 The Evolution Towards WiMAX

39

Equipment Submission Vendor to: Select certification profile and optimal modules Participate in plugfests and conduct self testing

Conformance Testing Certification Lab to: Test Protocol Compliance Test Radio Compliance

No

Pass

Yes Interoperability Testing Certification Lab to: Test Interoperability with products from other vendors

No

Pass

Yes Certification Issue WiMAX Forum to: Announce certification Issue Certificate Issue test report to vendor

Fig. 1.18 Plugfest algorithm (From WiMAX forum)

l

WiMAX equipment. The list of features tested in system profiles is more stringent than the underlying standards (features that are optional in the standards may be tested as mandatory by the WiMAX Forum Certified program), but does not include any new feature that is not included in the standards. For instance, the Fixed WiMAX profile is based on IEEE 802.16-2004 and only allows testing on equipment using point to multipoint operations up to 11 GHz, while IEEE 802.16-2004 equipment can operate up to 66 GHz, Table 1.16. Similarly, Fixed WiMAX uses OFDM multiplexing with 256 carriers, even though IEEE 802.16-2004 also supports an Orthogonal Frequency Division Multiple Access (OFDMA) mode. Mobile WiMAX is a second system profile, Table 1.17. The WiMAX Forum defines a list of test cases to use during the certification process for all equipment based on the same system profile. Certification profiles define classes of interoperable equipment for the testing process. They are instances of a system profile, defined by three parameters:

40

R. Prasad and F.J. Velez

Table 1.16 Fixed WiMAX certification profiles

Profile name ET01 ET02

Spectrum band (GHz) 3.4–3.6 3.4–3.6

Channel Duplexing Status bandwidth (MHz) 3.5 TDD Active 3.5 FDD Active

Table 1.17 Mobile WiMAX certification profiles [26] Profile name Spectrum band (GHz) Channel bandwidth (MHz) MP01 2.3–2.4 8.75 MP02 2.3–2.4 5 and 10 MP05 2.496–2.69 5 and 10 MP09 3.4–3.6 5 MP10 3.4–3.6 7 MP12 3.4–3.6 10 a Projected start date of certification testing, subject to change

l l l

Duplexing TDD TDD TDD TDD TDD TDD

Status Active 2009a Active 4Q2008a 4Q2008a 4Q2008a

Spectrum band ( 2 Mbps Media content download Bulk data, movie download > 1 Mbps (store and forward) Peer-to-peer > 500 kbps

generation of wireless systems is relatively expensive as without a standard few economies of scale are possible. These limitations of traditional wired and proprietary wireless technologies can be overcome by the absence of LoS requirement, high bandwidth, inherent flexibility and range of IEEE 802.16 solutions. Therefore, fixed/nomadic WiMAX is ideally suited to be the best alternative to both broadband internet and voice telephony. The family of IEEE 802.16 standards effectively addresses several different types of potential customers and situations, especially where alternative DSL solutions are neither available nor economically viable. It is expected that standards based solutions will enable the largest and most flexible available ecosystem, driving competition and features up. Furthermore, this will bring prices down enabling to create new applications and services. Its ability to support both LoS and NLoS connections make WiMAX suitable for ubiquitous service offering in rural and urban areas alike. High speed and symmetrical bandwidth may satisfy the needs of individual customers, public administrations, and enterprises of all sizes. Cellular coverage makes its deployment extremely fast and relatively inexpensive. There is a vast number of real time services and applications that are already provided on WiMAX networks. This means that service providers are not forced to develop special applications specifically for WiMAX. Few of such applications, which already exist on wired networks are classified under service classes and listed in Table 1.22. The high bandwidth provided by WiMAX facilitates to deliver real-time applications much faster than other technologies do today. Thus, in terms of applications, WiMAX means freedom to both subscribers and service providers. Mobile WiMAX (IEEE 802.16e) will also encourage the early emergence of mobile applications that address the specific needs of mobile Internet users. IMS support in WiMAX will facilitate the ubiquitous deployment of managed services for subscribers across wired and wireless platforms.

1.13.1 Fixed Broadband Wireless Applications The market for Fixed Broadband Wireless Access (FBWA) is huge. New and lucrative broadband access markets are springing up everywhere while new carriers

50

R. Prasad and F.J. Velez

and service providers are being created. In emerging countries, Plain Ordinary Telephone System (POTS) penetration is still low. Also, the low quality of the copper pair prevents mass scale DSL deployment fostering the need for alternate broadband technologies. FBWA fills the void at markets where cable or copper infrastructures are either saturated, outdated or simply out of reach, providing highly efficient and cost effective access services for millions of subscribers who would otherwise be left out of the loop. The BWA market targets wireless multimedia services to Small Offices/Home Offices (SOHOs), small and medium-sized businesses, and residences. The fixed wireless market for broadband megabit per second transmission rates is growing for providing an easily deployable low-cost solution, compared with existing cable and digital subscriber line (xDSL) technologies, for dense and suburban environments.

1.13.2 VoIP over WiMAX Voice over IP (VoIP) is expected to be one of the dominating WiMAX services in future. VoIP over WiMAX is attractive both for enterprises and carriers. In VoIP, voice is transferred as data packets through personal broadband networks or public Internet. This makes VoIP service provider bypass the high operation costs for PSTN by using the cheapest resource of Internet. Although WiMAX is not designed for switched cellular voice traffic as cellular technologies, like CDMA and WCDMA, its rich set of QoS classes, supporting low latency, provide full support for VoIP traffic. VoIP calls can be received or placed at a very low or, in some cases, no additional cost. Therefore, the VoIP over WiMAX solution is the combination between the wireless technology and the low price VoIP. WiMAX is not a technology specifically built for voice communications; therefore, it will not challenge the voice revenues of mobile operators. This is mainly because cellular networks offer a cost-effective infrastructure for voice communications, with an extensive coverage that WiMAX is not designed to replace. But in some cases a mobile operator may move some voice traffic to the WiMAX infrastructure due to capacity constraints. Other real-time applications, like mobile video and audio streaming, videoconferencing and gaming, will strongly benefit from QoS support and low latency. They will become increasingly important when new devices optimized for these applications are introduced. Broadcast is another potential WiMAX application. Work is currently under way within the WiMAX Forum to further develop the Multicast and Broadcast Services (MBS) protocols within the standard, to allow for efficient multicasting of content. Vertical applications like surveillance, public safety, connectivity to remote devices, inventory tracking, fleet management and educational services can also

1 The Evolution Towards WiMAX

51

FRACTIONAL E1 for SMALL BUSINESS BACKHAUL for HOTSPOTS

T1+ LEVEL SERVICE ENTERPRISE

Mobile Backhaul

RESIDENTIAL & SoHo DSL LEVEL SERVICE

802.16d WMAN Nomadic Coverage --> handoff from HOT SPOTS H

H

H

H

802.16e

H

H H

H H

= wide area coverage outside of Hot Spots

INTERNET BACKBONE BWA Operator Network Backbone

Fig. 1.21 WiMAX applications and missions [35]

be supported by mobile WiMAX networks with little or no incremental cost to network operators. These applications require robust and reliable connectivity, but in most of the cases it would be prohibitively expensive to build separate networks to support them. A WiMAX operator is thus well placed to support these applications and to secure new revenue streams either by providing the service to new market segments or by establishing wholesale relationships with service providers that focus on specific verticals. In fact, WiMAX is ideally suited for the following the following applications and missions, Fig. 1.21: l l

l l

l

l

Extending DSL/cable modem services to rural areas Backhauling traffic for wireless service providers or cable operators at a reduced cost Interconnecting and backhauling Wi-Fi hot zones Enabling ISPs, cable and satellite operators to deliver existing content through a new channel, which is even more attractive with WiMAX 802.16e Providing robust, secure bandwidth for data traffic at widely dispersed enterprises, such as financial and educational institutions and municipalities Enabling secure communications in-the-field for the military or public safety institutes

52

R. Prasad and F.J. Velez

The IEEE 802.16 standard can help the industry to provide solutions across multiple broadband segments, namely, (1) cellular backhaul, (2) broadband ondemand, (3) residential broadband, and (4) underserved areas [35]. The robust high bandwidth support of IEEE 802.16 technology makes it also a good choice for commercial enterprises backhaul. The on-demand connectivity is useful in trade shows and also in businesses such as construction sites. The deployment of IEEE 802.11 hotspots and home/small office wireless LANs in areas not served by cable or DSL will be accelerated by last-mile broadband wireless access provided by WiMAX. For many businesses, the broadband connectivity is so much important that relocation of organizations to areas where service is available is done. If the service is not available, a local exchange carrier may take 3 months or more to provide a T1 line to business customer. IEEE 802.16 wireless technologies enable to provide high speed services (with a speed comparable to wired services) at low cost, with a response time of only some days. The service providers with IEEE 802.11 wireless technologies also offer instantly configurable on-demand high-speed connectivity for temporary events, generating hundreds or thousands of users through 802.11 hotspots.

1.13.2.1

Wi-Fi Backhaul

Hotspots are becoming a very popular means of surfing on public places like coffee shops, airports, train stations, etc. These hotspots continue to be set up on worldwide basis at a very fast rate. For every hotspot, there is a DSL or dedicated broadband medium which is used to backhaul its traffic from the wireless access point to the telephone company central office, enabling to connect to the Internet. For these hotspots WiMAX may therefore serve as high bandwidth Internet backhaul. Another area where WiMAX connectivity may be very useful is Wi-Fi hot spots connectivity. There may be several Wi-Fi hotspots whose WiMAX backhaul provides a full wireless solution. Thus, last-mile broadband wireless access will help to accelerate the deployment of IEEE 802.11 hotspots and home/small office wireless LANs, especially in those areas not served by cable or DSL or in areas where the local telephone company may have a long lead time for provisioning broadband service.

1.13.2.2

Cellular Backhaul

Backhaul for cell towers is currently enabled through either a dedicated T1/E1 medium or a microwave radio link but WiMAX may be a better alternative for these types of backbone. Also, given the limited availability of a wired infrastructure in some developing countries, the costs to install a WiMAX station in conjunction with an existing cellular tower (or even as a solitary hub) are likely to be lower in comparison to developing a wired solution. Areas of low population density and flat

1 The Evolution Towards WiMAX

53

terrain are particularly suited to WiMAX, whose coverage range is appropriate. For countries that have skipped wired infrastructure as a result of prohibitive costs and irregular geography, WiMAX can enhance the Wireless infrastructure in an inexpensive and decentralized deployment – in a friendly and effective manner. There are number of advantages such as reduced operating expense (OPEX). Besides, no monthly rent for fixed line service need to paid as well as a new network is made available supporting broadband roaming.

1.13.2.3

Rural Connectivity

Rural areas are typically small city or towns that are located far from metropolitan area. Such areas are underserved and do not have enough wired broadband connection. For small businesses in remote areas, such as rural towns and mountain communities, high-speed Internet access has been hard to come by. The conventional broadband services providers, like DSL and cable companies, have not typically extended their services outside of well-to-do cities and suburbs because such a build-out may be very expensive. However, being off-the-Internet is not a plausible possibility for any business these days. High-speed Internet access is essential to gather information, and communicate with partners, clients and potential customers. The companies also need to maintain a presence on the Web in today’s global marketplace. Rural connectivity is therefore critical in many developing countries and underserved areas of developed country. Rich WiMAX QoS features ensure the real time voice transmission as well as low latency. Rural connectivity is therefore essential for voice telephony and Internet services. As the WiMAX solution provides extended coverage, it is a cost effective solution, and it can be preferable to wired technology in areas with low population density. Furthermore, WiMAX solutions may be quickly deployed.

1.13.2.4

Last Mile to the Home

The residential market segment primarily depends on the availability of DSL or cable access, which are however poor and expensive to build up connectivity in the rural or suburban areas. The lowest cost and highest capacity of the WiMAX solution will replace these conventional wired broadband technologies. Thus, low cost WiMAX solutions may replace cable and ADSL solutions while really making the broadband home revolution happen.

1.13.2.5

Business Users

Worldwide, only 5% of commercial structures are served by fibre networks, the main method for the largest enterprises to access broadband multimedia data services. These networks are extended to the business or residence via cable or

54

R. Prasad and F.J. Velez

DSL. Both are expensive options because of the infrastructure changes required. Enterprises can use WiMAX instead of T1 for about 10% of the cost, while SMEs can be offered fractional T1 services. Base stations will cost under $20,000 and support 60 enterprise customers with T1-class connections.

1.13.2.6

Municipal and Law Enforcement

A number of local municipalities have already started city-wide implementation of wireless networks or at least they have announced plans to do so. The city councils want to make affordable ubiquitous broadband access available to residences and small medium enterprises (SMEs). In many cases, Wi-Fi would be the natural choice, primarily for its low cost and wide availability, including places as public libraries, schools, parks, and municipal buildings. WiMAX will play a key role in backhaul or backbone access to these Wi-Fi hot spots while providing the last mile link to SMEs. Additionally, WiMAX is an excellent technology with the following envisaged application scenarios: l l l l l l

Public safety Traffic management Parking meter reading Public utilities Remote utility reading Toll collection

1.13.2.7

Emergency Response and Monitoring

During national crisis, wireless communications play a critical role. For example, in fire emergency, WiMAX can play a critical role by providing mobile transmission of maps, floor layouts and architectural drawings. This can assist fire-fighter brigades and other response personnel in the rescue of individuals involved in emergency situations. In situations of national crisis, such as terrorist attacks or natural disaster (e.g., hurricanes, earthquakes and floods), where the terrestrial infrastructure is broken and speed of deployment is of utmost importance, IEEE 802.16 can provide quick relief. In addition to disaster relief, WiMAX could be used in law enforcement applications. Examples are the following such as: l

l l l

Video/sensor surveillance: Wireless video surveillance is a cost-effective, flexible and reliable tool for monitoring traffic, roads, bridges, dams, offshore oil and gas, military installations, perimeter, borders and many more critical locations Public safety City workforce mobility Real-time video-based criminal checks

1 The Evolution Towards WiMAX

55

WiMAX may also be used in remote monitoring of patients to provide continuous information and immediate response in the event of a patient crisis.

1.14

Salient Features of WiMAX

WiMAX is a wireless broadband solution that offers a rich set of features, with a lot of flexibility in terms of deployment options and diversity of service offerings. Some of the most salient WiMAX features are the following [20, 29]: l

l

l

l

OFDM-based PHY layer – The WiMAX PHY layer is based on OFDM, a scheme that uses many subcarriers, or tones, to carry a signal. OFDM signalling consists of a large set of spaced subcarriers with no mutual interface to perform parallel data transmission in the frequency domain. The primary advantage of OFDM is that it eliminates the multi-path channel-induced self-interference that occurs with high speed data transmission over the wireless channel, which can limit transmission speed and spectral efficiency. OFDM offers good resistance to multipath, and allows WiMAX to operate in NLOS conditions. Very high peak data rates – The peak data rate indicates the bit-rate a user in good radio conditions can reach when not sharing the channel with other users. WiMAX is capable of supporting very high peak data rates. In fact, the peak PHY data rate can be as high as 70 Mbps when operating using a 20 MHz wide bandwidth. More typically, using a 10 MHz spectrum operating using TDD scheme with a 3:1 downlink-to-uplink ratio, the peak PHY data rate is about 25 Mbps and 6.7 Mbps for the downlink and the uplink, respectively. These peak PHY data rates are achieved when using 64 QAM modulation with rate 5/6 error-correction coding. Under very good signal conditions, even higher peak rates may be achieved using multiple antennas and spatial multiplexing. Scalable bandwidth and data rate support – WiMAX has a scalable PHY layer architecture that allows for the data rate to scale easily with available channel bandwidth. This scalability is supported in the OFDMA mode, where the FFT (fast Fourier transform) size may be scaled based on the available channel bandwidth. For example, a WiMAX system may use 128-, 512-, or 1,048-bit FFTs based on whether the channel bandwidth is 1.25 MHz, 5 MHz, or 10 MHz, respectively. This scaling may be done dynamically to support user roaming across different networks that may have different bandwidth allocations. Adaptive modulation – During the time of transmission the quality of the radio link always varies. To overcome the transmission challenges, WiMAX supports a number of modulation schemes. Adaptive modulation allows the WiMAX system to adjust the signal modulation scheme depending on the signal to noise ratio (SNR) of the channel conditions. When the radio link is high in quality, the highest modulation scheme is used increasing the system capacity. The system shifts to a lower modulation scheme during a signal fade. This increases the

56

l

l

l

l

l

R. Prasad and F.J. Velez

range, as higher modulation scheme is used over as opposed to having a fixed scheme for worst channel conditions. Error Correction Techniques – WiMAX incorporates strong error correction techniques like FEC and ARQ to improve throughput. In FEC errors are detected and corrected upon reception by adding redundancy to the transmitted signal. Reed Solomon FEC, convolutional coding and interleaving algorithms are used in WiMAX systems. ARQ is a MAC level error control scheme and is used to correct errors not corrected by FEC. In ARQ scheme the information with error is resent. WiMAX systems also support hybrid technique, called H-ARQ, where a combination of FEC and ARQ is used. Support for TDD and FDD – IEEE 802.16-2004 and IEEE 802.16e-2005 supports both FDD (Frequency division duplexing) and TDD (Time division duplexing), as well as a half-duplex FDD. FDD is the legacy duplexing method that has been deployed in cellular telephony. In regulatory environments where structured channel pairs do not exist, TDD uses a single channel for both upstream and downstream transmissions, dynamically allocating bandwidth depending on traffic requirements. TDD is also favoured by a majority of implementations because of its advantages like, ability to exploit channel reciprocity, ability to implement in non-paired spectrum, and less complex transceiver design. All the initial WiMAX profiles are based on TDD, except for two fixed WiMAX profiles in 3.5 GHz. Orthogonal frequency division multiple access (OFDMA) – Mobile WiMAX uses OFDM as a multiple-access technique, whereby different users can be allocated different subsets of the OFDM tones. As discussed in detail in Chapters 2 and 11, OFDMA facilitates the exploitation of frequency diversity and multiuser diversity to significantly improve the system capacity. Flexible and dynamic per user resource allocation – Both uplink and downlink resource allocation are controlled by a scheduler in the base station. Capacity is shared among multiple users on a demand basis, using a burst TDM scheme. When using the OFDMA-PHY mode, multiplexing is additionally done in the frequency dimension, by allocating different subsets of OFDM subcarriers to different users. Resources may be allocated in the spatial domain as well when using the optional advanced antenna systems (AAS). The standard enables bandwidth resources to be allocated in time, frequency, and space and has a flexible mechanism to convey the resource allocation information on a frame-by-frame basis. Support for advanced antenna techniques – Advanced multi-antenna technologies are one of the most important methods of improving spectral efficiency in non-cellular wireless networks. 802.16 standards allow vendors to support a variety of these mechanisms, which can be a key performance differentiator. The WiMAX solution has a number of built in features into the physical-layer design, which allows for the use of multiple-antenna techniques, such as beamforming, space-time coding, and spatial multiplexing. These schemes can be used to improve the overall system capacity and spectral efficiency by deploying

1 The Evolution Towards WiMAX

l

l

l

l

57

multiple antennas at the transmitter and/or the receiver. Chapter 11 presents detailed overview of the various multiple antenna techniques. Quality-of-service support – WiMAX system supports a multiplicity of widely varying transport demands over an inherently fluctuating wireless medium. Therefore the network must include rigorous support for differentiated QoS as a fundamental design feature. WiMAX MAC layer has a connection-oriented architecture that is designed to support a variety of applications, including voice and multimedia services. The grant/request mechanism enables an operator to simultaneously provide premium guaranteed levels of service to businesses, such as T1-level service, and high-volume “best-effort” service to homes, similar to cable-level service, all within the same base station service area cell. Also the MAC is designed to support a large number of users, with multiple connections per terminal, each with its own QoS requirement. The system offers support for constant bit rate, variable bit rate, real-time, and non-real-time traffic flows, in addition to best-effort data traffic. Robust security – WiMAX supports strong encryption, using Advanced Encryption Standard (AES), and has a robust privacy and key-management protocol. The system also offers a very flexible authentication architecture based on Extensible Authentication Protocol (EAP), which allows for a variety of user credentials, including username/password, digital certificates, and smart cards. Support for mobility – The mobile WiMAX variant of the system has mechanisms to support secure seamless handovers for delay-tolerant full-mobility applications, such as VoIP. The system also has built-in support for powersaving mechanisms that extend the battery life of handheld subscriber devices. Physical-layer enhancements, such as more frequent channel estimation, uplink subchannelization, and power control, are also specified in support of mobile applications. IP-based architecture – Network Working Group (NWG) in the WiMAX Forum has defined a reference network architecture that is based on an all-IP platform. WiMAX has been at the forefront of the move to all-IP end-to-end networks based on open systems. All end-to-end services are delivered over an IP architecture relying on IP-based protocols for end-to-end transport, radio resource management, QoS management, security, and mobility. Service providers can concentrate on a core packet based network and have services that enable them provide voice, WLAN and broadcast services over WiMAX. Reliance on all-IP allows WiMAX systems to facilitate easy convergence with other networks and enter the rich ecosystem for application development that exists for IP.

1.15

Fixed, Portable, and Mobile Terminals

The term Wireless does not exactly mean mobile. There is therefore often a need to make a conceptual difference between the terms ‘Wireless’, ‘Portable’ and ‘Mobile’. Wireless systems may be classified into two types: Mobile and Fixed.

58

R. Prasad and F.J. Velez

Fixed wireless system include traditional broadcast television, direct to home satellite service and the microwave backhaul networks that transport telephone traffic from individual cell sites to the mobile switching centre. In these systems the transmitters and receivers are permanently fixed in one geographical location. For example, PtP microwave links are permanently located and aimed so that each radio station is in LoS with the corresponding ones. If the sites are moved or misaligned, then the radio contact is lost. In the Wireless context, the term fixed is clear: the transmitting and receiving terminals of the wireless transmission circuit are physically fixed in a place. A microwave link system, with the transmitting and receiving antennas mounted on towers attached to the ground, a rooftop or to some other structure is a reference example of a fixed Wireless system. Television broadcast systems are fixed systems in this sense. The transmitting and receiving antennas are fixed, as is the TV itself while it is being watched. An exception to this are high-gain antennas receiving satellite signals on board ships where sophisticated gimbaling systems are required to keep the antenna pointed in the correct orientation regardless of the movements of the ship. The IEEE 802.11 standards are primarily designed for fixed and ‘portable’ terminal devices. Also referred to as nomadic in the context of ITU and other European standardization bodies, a portable terminal is one that stays in one place while it is being used but can readily be picked up and moved to another location. A notebook computer, with an 802.11b wireless access PC card, is a good example of such a portable device. Another portable device is a desktop wireless modem that is connected to a computer via a USB port. While operating, the modem is expected to steadily be in one place – on a desktop, for example. Moving it across the desk or to another room makes it portable but while it is in use it is stationary. The classification of fixed systems can be further refined by recognizing that for some networks, one terminal of the transmission link is at an ad-hoc location rather than an ‘engineered’ (or ‘planned’) location; that is, no planning effort has been made to optimize the location for the terminal device. Instead, the terminal location has been chosen by the user as the most convenient one. A notebook computer with a wireless modem of some sort, placed on an arbitrarily located desk, is an example of such a system. Ad-hoc fixed systems present new challenges to system design since the problem of analyzing coverage and interference, and ultimately performance is quite similar to that of mobile radio or cellular systems in which the design must provide for terminals located essentially anywhere in the system service area. The differences between engineered and ad-hoc fixed wireless systems have an important impact on the commercial success of the system. An engineered system requires the expensive step of sending a trained engineer or technician to every terminal location at least once to complete a successful installation. The value to the operator of this customer’s business must be significant enough to justify the cost of this ‘truck roll’. Certainly, for some customers, such as large businesses that require microwave links carrying hundreds of megabits of data, this may very well be the case. However, for systems designed to serve thousands of more casual communication users, for example, at homes, home offices, and small businesses, a system

1 The Evolution Towards WiMAX Fig. 1.22 Mobile WiMAX enabled device can move from one Base Station (BS) to another in a seamless session [36]

59

BS

BS

BS PDA (WiMAX Mobile Ennabled)

BS

design that can work effectively with ad-hoc ‘self-installed’ terminals, without the necessity of truck roll, is essential for commercial viability. Mobile phones, pagers Family Radio Service (FRS) also called walkie-talkies and some satellite phones are the example of Mobile Wireless Systems. These radio systems are designed for situations where the handset is expected to be in motion while being used. As shown in Fig. 1.22, communication is not interrupted when Mobile WiMAX enabled device moves from the range of one Base Station (BS) to another while the user moves in seamless manner. With mobility there is no realtime speed limit. High mobility is accomplished when a Broadband Wireless Access (BWA) device can be used in some high speed trains with speeds exceeding 300 km/h, or more. WiMAX Ecosystem The WiMAX ecosystem is a common platform created by all industry players in theWiMAX value chain. For WiMAX to succeed, all the stakeholders of this ecosystem must come together to develop, produce, certify, promote and deliver a quality product. The WiMAX ecosystem consists of standardization bodies, operators, service providers, equipment vendors, system integrators, component vendors, and users, Fig. 1.23. The success of the BWA business depends on a stable and comprehensive industrial ecosystem. WiMAX forum offers a platform of cooperation for stakeholders to build up such ecosystem. This ecosystem is open and dynamic allowing its members to join in the process of value creation. Just at the chip level, the variety of companies delivering solutions ranging from the OFDM technology being used in Mobile WiMAX to a wide variety of systemon-a-chip silicon providers, offer the variety of support and implementation strategies that will either ensure or fail to ensure that the technology supports the range of optional features that will garner customer attention. The various equipment vendors as well as the component and system integrators deliver unique solutions tailored to the driving market needs. They offer opportunities to them and their service provider customers.

60

R. Prasad and F.J. Velez

Component Vendor

System Integrator

User Residential Enterprise Mobile

Operator

Service provider

WiMAX Forum

Equipment Vendor

Fig. 1.23 WiMAX ecosystem [30, 37]

The WiMAX ecosystem is a crucial element of success for this technology. The widespread quantity and quality of the companies supporting WiMAX in their various ecosystem roles is a major reason that the market place has taken the technology so seriously. It is also the industry’s greatest hope for success. Many famous vendors from different industry area have come together and joined hands for the development of WiMAX equipment and enhance interoperability for various products. WiMAX Forum promotes the development of WiMAX profile and undertakes the certification of the products to ensure the interoperability of equipment from different vendors. The end user, that is, the customer, is certainly being benefited from such a healthy ecosystem.

1.16

Conclusions

WiMAX is an emerging and exciting wireless technology that will support a variety of business and consumer applications, from network backhauling and interconnecting Wi-Fi and LANs, to voice, video, data, and mobility support. WiMAX will change the way people access data, e-mail, and instant messaging service. It may coexist with 3G or even replace 3G, and it can seamlessly integrate with cellular handsets or serve as a standalone technology to provide mobile voice and data services. Moreover, WiMAX technology can be used in a variety of commercial, municipal, and law enforcement business applications. The success of WiMAX deployment will depend on its adoption rate, equipment pricing, business models, availability and affordability of licensed spectrum. Based on Internet access is becoming a necessity, and 3G and Research in Motion (RIM) capturing a large business market share, the WiMAX market is continuously growing. It will change the way people access the Internet and

1 The Evolution Towards WiMAX

61

download media, such as streaming video and MP3 files. Successful deployments have so far been limited to a few niche applications and markets. Clearly, broadband wireless until now had not reached to the required record, in part because of the fragmentation of the industry due to the lack of a common standard. The emergence of WiMAX as an industry standard is expected to change this situation.

References 1. R. Prasad, Universal Wireless Personal Communications (Artech House, Norwood, MA, 1998) 2. A.R. Prasad, The future re-visited. Wireless Pers. Commun. Int. J. 37(3–4), 187–211 (May 2006) 3. R. Bekkers, Mobile Telecommunication Standards, GSM, UMTS, TETRA, ERM S (Artech House, Norwood, MA, 2001) 4. V.K. Garg, Wireless Network Evolution: 2G to 3G (Prentice Hall, Upper Saddle River, NJ, 2002) 5. Y.K. Kim, R. Prasad, 4G Roadmap and Emerging Communication Technologies (Artech House, Boston, MA, 2006) 6. R. Prasad, R.L. Olsen, The unpredictable future: personal networks paving towards 4G. Telektronikk 1.2006, Real-time communication over IP. 7. S. Ahamdi, An overview of next-generation mobile WiMAX technology. Commun. Mag. 47(6), 84–98 (June 2009) 8. R. Prasad, Convergence: a step towards unpredictable future (keynote speech), in Proceedings of EuWIT2009 – European Wireless Technology Conference 2009 (Rome, Italy, Sept 2009) 9. R. Prasad, M. Ruggier, Technology Trends in Wireless Communication (Artech House, Norwood, MA, 2003) 10. B. Fong, N. Ansari, A.C.M. Fong, G.Y. Hong, P.B. Rapajic, On the scalability of fixed broadband wireless access network deployment. IEEE Radio Communications (Sept 2004) 11. A. Ghosh, R. Muhamed, J.G. Andrews, Fundamentals of WiMAX: Understanding Broadband Wireless Networking (Prentice Hall, Upper Saddle River, NJ, 2007) 12. V. Gunasekaran, F.C. Harmantzis, Emerging wireless technologies for developing countries. Technol Soc 29(1), 23–42 (Jan 2007) 13. R.B. Marks, Status of 802.16 Efforts: successes, new products & next steps, in Wireless Communication Association 8th Annual Technical Symposium (Jan 2002) 14. IEEE 802.16 Working Group web site. http://WirelessMAN.org (March 2010) 15. F. Ohrtman, WiMAX Handbook: Building 802.16 Wireless Networks (McGraw-Hill, New York, 2005) 16. International Telecommunications Union (ITU). http://www.itu.int/net/home/index.aspx (March 2010) 17. European Telecommunication Standardization Institute (ETSI). http://www.etsi.org/ (March 2010) 18. Rules of the Federal Communication Commission. Part 101.101. CFR Title 47. United States Government Printing Office 19. Rules of the Federal Communication Commission. Part 74. CFR Title 47. United States Government Printing Office 20. C. Eklund, R. Marks, K. Stanwood, S.Wang, IEEE standard 802.16: a technical overview of the WirelessMANTM air interface for broadband wireless access

62

R. Prasad and F.J. Velez

21. IEEE 802.16a-2003, IEEE standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed Broadband Wireless Access Systems – Medium Access Control Modifications and Additional Physical Layer Specifications for 2–11 GHz (Jan 2003) 22. IEEE 802.16-2004, IEEE Standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed Broadband Wireless Access Systems (Oct 2004) 23. IEEE 802.16e-2005, IEEE Standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems Amendment 2: Physical and Medium Access Control Layers for Combined Fixed and Mobile Operation in Licensed Bands (Feb 2006) 24. S. Shepard, WIMAX Crash course (McGraw-Hill, New York, 2006) 25. D. Pareek, The Business of WiMAX (Wiley, Hoboken, NJ, 2006) 26. White Paper (The WiMAX Forum Certified TM Program). www.wimaxforum.org (March 2010) 27. IEEE 802.16c-2002, IEEE Standard for Local and Metropolitan Area Networks – Part16: Air Interface for Fixed Broadband Wireless Access Systems – Amendment1: Detailed System Profiles for 10-66 GHz (Dec 2002) 28. http://en.wikipedia.org/wiki/IEEE_802.16#cite_note-0 (March 2010) 29. Mobile WiMAX – Part I: A Technical Overview and Performance, WiMAX Forum (Mar 2006) 30. www.wimaxforum.org (March 2010) 31. Fujitsu Microelectronics America Inc., RF Spectrum Utilization in WiMAX, white paper (2004) 32. Maravedis Inc, Spectrum Analysis – The Critical Factor in BWA/WiMAX versus 3G (Jan 2006) 33. White Paper (WiMAX and IMT-2000). www.wimaxforum.org (March 2010) 34. WiMAX Forum (Mobile WiMAX – Part II: Competitive Analysis). www.wimaxforum.org (March 2010) 35. Can WiMAX Address Your Applications? Westech on Behalf of the WiMAX Forum (Oct 24, 2005) 36. WiMAX Forum, Fixed, nomadic, portable ad mobile applications for 802.16-2004 and 802.16e WiMAX networks (2005). www.wimaxforum.org (March 2010) 37. White Paper (M-Taiwan Program: A WiMAX Ecosystem). www.wimaxforum.org (March 2010)

Chapter 2

OFDMA WiMAX Physical Layer Ramjee Prasad and Fernando J. Velez

Abstract IEEE 802.16 physical (PHY) layer is characterized by Orthogonal Frequency Division Multiplexing (OFDM), Time Division Duplexing, Frequency division Duplexing, Quadrature Amplitude Modulation and Adaptive Antenna Systems. After discussing the basics of OFDM and Orthogonal Frequency division Multiple Access (OFDMA), scalable OFDMA is presented and supported frequency bands, channel bandwidth and the different IEEE 802.16 PHY are discussed. The similarities and differences between wireless MAN-SC, wireless MAN-OFDM and wireless MAN-OFDMA PHY are finally highlighted.

2.1

Introduction

The IEEE 802.16 standard belongs to the IEEE 802 family, which applies to Ethernet. WiMAX is a form of wireless Ethernet and therefore the whole standard is based on the Open Systems Interconnections (OSI) reference model. In the context of the OSI model, the lowest layer is the physical layer. It specifies the frequency band, the modulation scheme, error-correction techniques, synchronization between transmitter and receiver, data rate and the multiplexing techniques. For IEEE 802.16, Physical layer was defined for a wide range of frequencies from 2–66 GHz. In sub frequency range of 10–66 GHz there essentially is LoS propagation. Therefore, single carrier modulation was chosen, because of low

R. Prasad (*) Center for TeleInFrastruktur (CTIF), Aalborg University, Niels Jernes Vej 12, DK–9220 Aalborg Øst, Denmark e-mail: [email protected]

R. Prasad and F.J. Velez, WiMAX Networks, DOI 10.1007/978-90-481-8752-2_2, # Springer ScienceþBusiness Media B.V. 2010

63

64

R. Prasad and F.J. Velez

system complexity. Downlink channel is shared among users with TDM signals. Subscriber unit are being allocated individual time slots. Access in uplink is being realized with TDMA. In the 2–11 GHz bands, communications can be achieved for licensed and non-licensed bands. The communication is also available in NLoS conditions. To ensure the most efficient delivery in terms of bandwidth and available frequency spectrum, the IEEE 802.16 physical layer uses a number of legacy technologies. These technologies include Orthogonal Frequency Division Multiplexing (OFDM), Time Division Duplexing (TDD), Frequency Division Duplexing (FDD), Quadrature Amplitude Modulation (QAM), and Adaptive Antenna System (AAS). The WiMAX physical layer is based on OFDM. OFDM is the transmission scheme of choice to enable high speed data, video, and multimedia communications and presently, besides WiMAX, it is used by a variety of commercial broadband systems, including DSL, Wi-Fi, Digital Video Broadcast-Handheld (DVB-H). Above the physical layer are the functions associated with providing service to subscribers. These functions include transmitting data in frames and controlling access to the shared wireless medium, and are grouped into a media access control (MAC) layer. This Chapter is organized as follows. Section 2.2 addresses the history, evolution and applications of OFDM. Section 2.3 presents the OFDM fundamentals by comparing it with FDMA as well as describing OFDM signal characteristics. Section 2.4 describes the concepts behind OFDM transmission and presents the serial to parallel converter as well as the OFDM demodulator. The OFDM symbol is described in Section 2.5 while Section 2.6 presents ISI and ICI mitigation. Section 2.7 addresses spectral efficiency. Section 2.8 presents the improvements of OFDMA and the advantages of subchannelisation. Section 2.9 and 2.10 present the advantages and disadvantages of OFDM systems. The details on scalable OFDMA are presented in Section 2.11, including the parameters, principles, and the reference model. Section 2.12 addresses specific issues of PHY layer, including WirelessHUMAN PHY, while Section 2.13 addresses WirelessMANSC (single carrier) PHY. Section 2.14 covers WirelessMAN-OFDM PHY while Section 2.15 describes WirelessMAN-OFDMA PHY. Finally, Section 2.16 presents the conclusions.

2.2 2.2.1

History and Development of OFDM Evolution

OFDM has recently been gaining interest from telecommunications industry. It has been chosen for several current and communications systems all over the world. Nevertheless, OFDM had a long history of existence (Table 2.1). The first multichannel modulation systems appeared in the 1950s as frequency division multiplexed

2 OFDMA WiMAX Physical Layer

65

Table 2.1 A brief history of OFDM Dates OFDM Landmark Achieved 1966 Chang postulated the principle of transmitting messages simultaneously through a linear band limited channel without ICI and ISI [3]. This is considered the first official publication on multicarrier modulation. Earlier work on OFDM was Holsinger’s 1964 MIT dissertation [4] and some of Gallager’s early work on waterfilling [5]. 1967 Saltzberg observed that, in order to increase efficiency of parallel system, cross talk between adjacent channels should be reduced [6]. 1971 Weinstein and Ebert show that multicarrier modulation can be accomplished by using a DFT [7]. 1980 Peled and Ruiz introduced use of Cyclic Prefix (CP) or cyclic extension instead of guard spaces [8]. 1985 Cimini at Bell Labs identifies many of the key issues in OFDM transmission and does a proof-of-concept design [9]. 1990–1995 OFDM was exploited for wideband data communications over mobile radio FM radio, DSL, HDSL, ADSL and VDSL. First commercial use of OFDM in DAB and DVB. 1999 The IEEE 802.11 used OFDM at the physical layer. HiperLAN and HiperLAN/ 2 also adopted OFDM at the physical layer. 2002 The IEEE 802.16 committee released WMAN standard 802.16 based on OFDM. 2003 The IEEE 802.11 committee releases the 802.11g standard for operation in the 2.4 GHz band. The multiband OFDM standard for ultra wideband is developed. 2004 OFDM is candidate for IEEE 802.15.3a standard for wireless PAN (MB-OFDM) and IEEE 802.11n standard for next generation wireless LAN [33]. 2005 OFDMA is candidate for the 3GPP Long Term Evolution (LTE) air interface E-UTRA downlink [33]. 2007 The first complete LTE air interface implementation was demonstrated, including OFDM-MIMO, SC-FDMA and multi-user MIMO uplink [34]. 2008 Mobile WiMAX base stations and subscriber devices were first certified by WiMAX Forum.

military radio links. OFDM had been used by US military in several high frequency military systems, such as KINEPLEX, ANDEFT and KATHRYN [1, 2]. In December 1966, Robert W. Chang outlined first OFDM scheme. This was a theoretical way to transmit simultaneous data stream through linear band limited channel without Inter Symbol Interference (ISI) and Inter Carrier Interference (ICI). Chang obtained the first US patent on OFDM in 1970 [12]. Around the same time, Saltzberg performed an analysis of the performance of the OFDM system and concluded that the strategy should concentrate more on reducing cross talk between adjacent channels than on perfecting the channels [6]. Until this time, we needed a large number of subcarrier oscillators to perform parallel modulations and demodulations. This was the main reason why the OFDM technique has taken a long time to become a prominence. It was difficult to generate such a signal, and even harder to receive and demodulate the signal. The hardware solution, which makes use of multiple modulators and demodulators, was somewhat impractical for use in the civil systems. In the year 1971, Weinstein and Ebert used Discrete Fourier Transform (DFT) to perform baseband modulation and demodulation. The use of DFT eliminated the

66

R. Prasad and F.J. Velez

need for bank of subcarrier oscillators. These efforts paved the way for the way for easier, more useful and efficient implementation of the system. The availability of this technique, and the technology that allows it to be implemented on integrated circuits at a reasonable price, has permitted OFDM to be developed as far as it has. The process of transforming from the time domain representation to the frequency domain representation uses the Fourier transform itself, whereas the reverse process uses the inverse Fourier transform. All the proposals until this moment in time used guard spaces in frequency domain and a raised cosine windowing in time domain to combat ISI and ICI. Another milestone for OFDM history was when Peled and Ruiz introduced Cyclic Prefix (CP) or cyclic extension in 1980 [8]. This solved the problem of maintaining orthogonal characteristics of the transmitted signals at severe transmission conditions. The generic idea that they placed was to use cyclic extension of OFDM symbols instead of using empty guard spaces in frequency domain. This effectively turns the channel as performing cyclic convolution, which provides orthogonality over dispersive channels when CP is longer than the channel impulse response [1]. It is obvious that introducing CP causes loss of signal energy proportional to length of CP compared to symbol length but, in turn, it facilitates a zero ICI advantage which pays off. By this time, inclusion of FFT and CP in OFDM system and substantial advancements in Digital Signal Processing (DSP) technology made it an important part of telecommunications landscape. In the 1990s, OFDM was exploited for wideband data communications over mobile radio FM channels, High-bitrate Digital Subscriber Lines (HDSL at 1.6 Mbps), Asymmetric Digital Subscriber Lines (ADSL up to 6 Mbps) and Very-high-speed Digital Subscriber Lines (VDSL at 100 Mbps). The first commercial use of OFDM technology was made in Digital Audio Broadcasting (DAB).The development of DAB started in 1987 and was standardized in 1994. DAB services started in 1995 in UK and Sweden. The development of Digital Video Broadcasting (DVB) was started in 1993. DVB along with High-Definition TeleVision (HDTV) terrestrial broadcasting standard was published in 1995. At the dawn of the twentieth century, several Wireless Local Area Network (WLAN) standards adopted OFDM on their physical layers. Development of European WLAN standard HiperLAN started in 1995. HiperLAN/2 was defined in June 1999 which adopts OFDM in physical layer. OFDM technology is also well positioned to meet future needs for mobile packet data traffics. It is emerging as a popular solution for wireless LAN, and also for fixed broad-band access. OFDM has successfully replaced DSSS for 802.11a and 802.11g. Perhaps of even greater importance is the emergence of this technology as a competitor for future fourth Generations (4G) wireless systems. These systems, expected to emerge by the year 2010, promise to at last deliver on the wireless ‘Nirvana’ of anywhere, anytime, anything communications [14]. It is expected that OFDM will become the chosen technology in most wireless links worldwide [13] and it will certainly be implemented in 4G radio mobile systems.

2 OFDMA WiMAX Physical Layer Table 2.2 Wireless systems using OFDM [10] Application WMAN Technology OFDM Cell Radius 1–20 km Mobility High and low Freq Band 2–66 GHz Data Rate Few Mbps Deployment IEEE 802.16a, d, e, WiMAX, 3GPP-LTE

2.2.2

67

WLAN OFDM up to 300 m Low 2–5 GHz up to 100 Mbps IEEE 802.11a, g, HiperLAN2

WPAN OFDM few tens of meter very low 5–10 GHz up to 10 Mbps IEEE 802.15, Zig-Bee

Applications of OFDM

OFDM has been incorporated into four basic applications: (1) Digital Audio Broadcasting (DAB); (2) Digital Video Broadcasting (DVB), over the terrestrial network Digital Terrestrial Television Broadcasting (DTTB); (3) Magic WAND (Wireless ATM Network Demonstrator); and (4) IEEE 802.11a/HiperLAN2 and MMAC WLAN Standards. DAB and DVD were the first standards to use OFDM. Next Magic WAND was introduced, which demonstrated the viability of OFDM. Lastly, and most importantly, the most recent 5 GHz applications evolved which were the first to use OFDM in packet-based wireless communications. Few of the OFDM application and their details based on the type of wireless access technique are summarized in Table 2.2.

2.3 2.3.1

OFDM Fundamentals OFDM Versus FDM

Orthogonal Frequency Division Multiplexing is an advanced form of Frequency Division Multiplexing (FDM) where the frequencies multiplexed are orthogonal to each other and their spectra overlap with the neighbouring carriers. As shown in the Fig. 2.1 the subcarriers never overlap for FDM. In contrast to FDM, OFDM is based on the principle of overlapping orthogonal sub carriers. The spectral efficiency of OFDM system as compared to FDMA is depicted in the Fig. 2.2. The overlapping multicarrier technique can achieve superior bandwidth utilization. There is a huge difference between the conventional non-overlapping multicarrier techniques such as FDMA and the overlapping multicarrier technique such as OFDM. In frequency division multiplex system, many carriers are spaced apart. The signals are received using conventional filters and demodulators. In these receivers guard bands are introduced between each subcarriers resulting into reduced spectral efficiency. But in an OFDM system it is possible to arrange the carriers in such a

68

R. Prasad and F.J. Velez Ch.1

Ch.2

Ch.3

Ch.4

Ch.5

Ch.6

Ch.7

Ch.8

Ch.9

Ch.10

frequency

Conventional multicarrier techniques

Ch.2 Ch.4 Ch.6 Ch.8 Ch.10 Ch.1 Ch.3 Ch.5 Ch.7 Ch.9 Saving of bandwidth

50% bandwidth saving frequency

Orthogonal multicarrier techniques

Fig. 2.1 Concept of OFDM signal

N=1

N=2

B = 2R

N=3 B = 2R

B = 2R

FDMA

–R

R

f

–R

f

R

–R

B = 3/2R

B=2R

–R/3

R/3

2R

f

B = 4/3R

OFDM –R

R

f

–3R/4 –R/4 R/4 3R/4

f

–2R/3 –R/3 R/3 2R/3

f

Fig. 2.2 Spectrum efficiency of OFDM compared to FDMA

fashion that the sidebands of the individual subcarriers overlap and the signals are still received without adjacent carrier interference. The main advantage of this concept is that each of the radio streams experiences almost flat fading channel. In slowly fading channels the inter-symbol interference (ISI) and inter-carrier interference(ICI) is avoided with a small loss of energy using cyclic prefix. In order to assure a high spectral efficiency the subchannel waveforms must have overlapping transmit spectra. But to have overlapping spectra, subchannels need to

2 OFDMA WiMAX Physical Layer

69

Subcarrier Peaks

Magnitude

Frequency

Subcarrier Nulls

Fig. 2.3 Orthogonal subcarriers in multicarrier systems (OFDM)

be orthogonal. Orthogonality is a property that allows the signals to be perfectly transmitted over a common channel and detected without interference. Loss of orthogonality results in blurring between the transmitted signals and loss of information. For OFDM signals, the peak of one sub carrier coincides with the nulls of the other sub carriers. This is shown in Fig. 2.3. Thus there is no interference from other sub carriers at the peak of a desired sub carrier even though the sub carrier spectrums overlap. OFDM system avoids the loss in bandwidth efficiency prevalent in system using non orthogonal carrier set.

2.3.2

OFDM Signal Characteristics

An OFDM signal consists of N orthogonal subcarriers modulated by N parallel data streams, Fig. 2.4. The data symbols (dn,k ) are first assembled into a group of block size N and then modulated with complex exponential waveform {fk(t)}. After modulation they are transmitted simultaneously as transmitter data stream. The total continuous-time signal consisting of OFDM block is given by xðtÞ ¼

1 X n¼1

"

N 1 X k¼0

# dn; k ’k ðt  nTd Þ

(2.1)

70

R. Prasad and F.J. Velez 1 0.8

Relative Amplitude

0.6 0.4 0.2 0 –0.2 –0.4 –5

–4

–3

–2

–1

0

1

2

3

4

1

2

3

4

5

Samples 1 0.8

Relative Amplitude

0.6 0.4 0.2 0 –0.2 –0.4 –5

–4

–3

–2

–1

0

5

Sample Duration Fig. 2.4 Spectra for OFDM subcarriers

where, fk(t) represents each baseband subcarrier and is given by  ’k ðtÞ ¼

e j2pfk t t 2 ½0; Td  0 otherwise

(2.2)

2 OFDMA WiMAX Physical Layer

71

where dn,k is the symbol transmitted during nth timing interval using kth subcarrier, Td is the symbol duration, N is the number of OFDM subcarriers and fk is kth subcarrier frequency, which is calculated as fk ¼ fo þ Tkd ; k ¼ 0 . . . N  1. Note that f0 is the lowest frequency used.

2.4 2.4.1

OFDM Transmission Concept

The OFDM based communication systems transmit multiple data symbols simultaneously using orthogonal subcarriers .The principle behind the OFDM system is to decompose the high data stream of bandwidth W into N lower rate data streams and then to transmit them simultaneously over a large number of subcarriers. Value of N is kept sufficiently high to make the individual bandwidth (W/N) of subcarriers narrower than the coherence bandwidth (Bc) of the channel. The flat fading experienced by the individual subcarriers is compensated using single tap equalizers. These subcarriers are orthogonal to each other which allows for the overlapping of the subcarriers. The orthogonality ensures the separation of subcarriers at the receiver side. As compared to FDMA systems, which do not allow spectral overlapping of carriers, OFDM systems are more spectrally efficient. OFDM transmitter and receiver systems are described in Figs. 2.5 and 2.6. At the transmitter, the signal is defined in the frequency domain. Forward Error Control/Correction (FEC) coding and interleaving block is used to obtain the robustness needed to protect against burst errors. The modulator transforms the encoded blocks of bits into a vector of complex values, Fig. 2.7. Group of bits are mapped onto a modulation constellation producing a complex value and representing a modulated carrier. The amplitudes and phases of the carriers depend on the data to be transmitted. The data transitions are synchronized at the carriers, and may be processed together, symbol by symbol.

Complex data constellations

BITS

Error Correction coding and Interleaving

Symbol Mapping (data modulation)

Fig. 2.5 OFDM transmitter

Baseband transmitted signal

Pilot symbol insertion

Serial-toparallel

OFDM Modulation via FFT

CP

DAC

IQ Modulation and upconverter

RF

72

R. Prasad and F.J. Velez

Received Complex data constellations

Received Signal at Baseband

Data conversion and I/Q demodulation

Carrier Synchronization

ADC

CP

OFDM Demodulation via IFFT

Parallel -to Serial-

Channel Estimation based on Pilot symbol

Symbol demapping (data demodulation)

Error correction decoding and Deinterleaving

O/P Binary data

Time Synchronization

Fig. 2.6 OFDM receiver

dn,0

ejw0t

S

x(t)

dn,N–1

Fig. 2.7 OFDM modulator

ejwN–1t

As the OFDM carriers are spread over a frequency range, chances are there that some frequency selective attenuation occurs on a time varying basis. A deep fade on a particular frequency may cause the loss of data on that frequency for that given time, thus some of the subcarriers can be strongly attenuated and that will cause burst errors. In these situations, FEC in COFDM can fix the errors [15]. An OFDM system with addition of channel coding and interleaving is referred to as Coded OFDM (COFDM). An efficient FEC coding in flat fading situations leads to a very high coding gain. In a single carrier modulation, if such a deep fade occurs, too many consecutive symbols may be lost and FEC may not be too effective in recovering the lost data. In a digital domain, binary input data is collected and FEC coded with schemes such as convolutional codes. The coded bit stream is interleaved to obtain diversity gain. Afterwards, a group of channel coded bits are gathered together (1 for BPSK, 2 for QPSK, 4 for QPSK, etc.) and mapped to the corresponding constellation points.

2 OFDMA WiMAX Physical Layer

2.4.2

73

Serial to Parallel Converter

Data to be transmitted is typically in the form of a serial data stream. Serial to parallel conversion block is needed to convert the input serial bit stream to the data to be transmitted in each OFDM symbol. The data allocated to each symbol depends on the modulation scheme used and the number of subcarriers. For example, in case a subcarrier modulation of 16-QAM each subcarrier carries 4 bits of data, and so for a transmission using 100 subcarriers the number of bits per symbol would be 400. During symbol mapping the input data is converted into complex value constellation points, according to a given constellation. Typical constellations for wireless applications are, BPSK, QAM, and 16 QAM. The amount of data transmitted on each subcarrier depends on the constellation. Channel condition is the deciding factor for the type of constellation to be used. In a channel with high interference a small constellation like BPSK is favourable as the required signal-to-noise-ratio (SNR) in the receiver is low. For interference free channel a larger constellation is more beneficial due to the higher bit rate. Known pilot symbols mapped with known mapping schemes can be inserted at this moment. Cyclic prefix is inserted in every block of data according to the system specification and the data is multiplexed to a serial fashion. At this point of time, the data is OFDM modulated and ready to be transmitted. A Digital-to-Analogue Converter (DAC) is used to transform the time domain digital data to time domain analogue data. RF modulation is performed and the signal is up-converted to transmission frequency. After the transmission of OFDM signal from the transmitter antenna, the signals go through all the anomaly and hostility of wireless channel. After the receiving the signal, the receiver downconverts the signal; and converts to digital domain using Analogue-to-Digital Converter (ADC). At the time of down-conversion of received signal, carrier frequency synchronization is performed. After ADC conversion, symbol timing synchronization is achieved. An FFT block is used to demodulate the OFDM signal. After that, channel estimation is performed using the demodulated pilots. Using the estimations, the complex received data is obtained which are de-mapped according to the transmission constellation diagram. At this moment, FEC decoding and deinterleaving are used to recover the originally transmitted bit stream. OFDM is tolerant to multi path interference. A high peak data rate can be achieved by using higher order modulations, such as 16 QAM and 64 QAM, which improve the spectral efficiency of the system.

2.4.3

Demodulator

The OFDM demodulator is shown in the form of a simplified block diagram is shown in Fig. 2.8. The orthogonality condition of the signals is based orthogonality of subcarriers {fk(t)}, defined by:

74

R. Prasad and F.J. Velez Td

∫ ( •)

dn,0

Td

e —jw0t x(t) Td

∫ (•)

dn,N–1

Td

e —jNw0t Fig. 2.8 OFDM demodulator

Z

’k ðtÞ’l ðtÞdt ¼ Td dðk  lÞ ¼



Td k ¼ l 0 otherwise

(2.3)


BS] Privacy Key Management Response [BS -> SS]

Management Message Type Code PKM Identifier (8 bits) (8 bits) (=9,for PKM-REQ)

TLV encoded Information

Fig. 5.6 PKM-REQ MAC management format

Management Message Type Code PKM Identifier (8 bits) (8 bits) (=10,for PKM-RSP)

TLV encoded Information

Fig. 5.7 PKM-RSP MAC management format

These MAC management message types distinguish between PKM requests (SS-to-BS) and PKM responses (BS-to-SS) as follows: 1. PKM Request (PKM-REQ) – The PKM-REQ message is sent from the SS to the BS, Fig. 5.6. It encapsulates one PKM message in its message payload. 2. PKM Response (PKM-RSP) – The PKM-RSP message is sent from the BS to SS, Fig. 5.7. It encapsulates one PKM message in its message payload. Each message encapsulates one PKM message in the Management Message Payload The X.509 certificates The X.509 certificates are used to identify the communicating parties. The IEEE 802.16 standard states that 802.16-compliant SSs must use X.509 Version 3 certificate formats, providing a public key infrastructure used for secure authentication. Each SS carries a unique X.509 digital certificate issued by the SS manufacturer, known as the SS X.509 certificate. This certificate is issued and signed by a Certification Authority (CA) and installed by the manufacturer. The digital certificate contains the SS RSA public key and the MAC address of the SS. IEEE 802.16 requires the X.509 certificate to come up with the following fields [1, 11]: l

l l l l l

The X.509 certificate version is always set to v3 when used in the 802.16 standard Certificate serial number Certificate issuer’s signature algorithm Public Key Cryptography Standard 1, that is, RSA encryption with SHA1 hashing Certificate issuer Certificate validity period

226 l

l

l

l

R. Prasad and F.J. Velez

Certificate subject, that is, the certificate holder’s identity, which, if the subject is the SS, includes the station’s MAC address Subject’s public key, which provides the certificate holder’s public key, identifies how the public key is used, and is restricted to RSA encryption Signature algorithm, which is identical to the certificate issuer’s signature algorithm Issuer’s signature, which is the digital signature of the Abstract Syntax

The standard works with two types of certificate: manufacturer and SS ones. It does not refer to BS certificates. Manufacturer certificates are used to identify the manufacturer of an IEEE 802.16 device. It could either be self-signed or issued by a third party. SS certificates are used for a single SS to be identified and its MAC address is present in the subject field. Manufacturers usually issue and sign the SS certificates. Then, BS is using the manufacturer’s public key and verifies the SS certificate through PKI. Table 5.5 shows the IEEE 802.16 encryption keys. Figure 5.8 presents the privacy and key management protocol. Table 5.6 presents the terms used in a PKM protocol message exchange. Table 5.5 IEEE 802.16 encryption keys Encryption Key Notation Number of bits Authorization AK 160 Key Key Encryption KEK 128 Key Traffic TEK 128 Encryption Key HMAC Key for HMAC_KEY_D 160 the Downlink HMAC Key for HMAC_KEY_U 160 the Uplink HMAC Key for HMAC_KEY_S 160 the Uplink

Generated Description by BS Authentication of an SS by its BS BS, SS BS

3DES key used for the encryption of the TEK Used for encrypting data traffic

Used for authenticating messages in the downlink direction. Used for authenticating messages in the uplink direction Used for authenticating messages in the Mesh mode

Message 1: BS ® SS: SeqNo | SAID | HMAC(1) Message 2: Key Request SS ® BS: SeqNo | SAID | HMAC(2) Message 3: Key Reply BS ® SS: SeqNo | SAID | OldTEK | NewTEK | HMAC(3)

Fig. 5.8 Privacy and key management protocol

5 Security Sublayer

227

Table 5.6 Terms used in a PKM protocol message exchange Term Description SeqNo The AK used for the exchange SAID The ID of the data SA being created or rekeyed HMAC(1) The HMAC-SHA1 digest of SeqNo | SAID under AK’s downlink HMAC key HMAC(2) The HMAC-SHA1 digest of SeqNo | SAID under AK’s uplink HMAC key OldTEK The previous-generation TEK’s initialisation vector, remaining lifetime (in seconds), and sequence number for the data SA specified by SAID (the TEK sequence number is a 2-bit quantity) NewTEK The next TEK’s initialisation vector, lifetime (in seconds), and sequence number for the data SA specified by SAID (the TEK sequence number is 1 greater, modulo 4, than the OldTEK sequence number) HMAC(3) The HMAC-SHA1 digest of SeqNo | SAID | OldTEK | NewTEK under AK’s downlink HMAC key

5.5.1

PKM Version 1

5.5.1.1

Security Associations

A Security Association (SA) is defined as the set of security information a BS and one or more of its client SSs share in order to support secure communications across the WiMAX access network. Three types of SAs have been defined, Primary, Static, and Dynamic. Each SS establishes a primary security association during the SS initialization process. Static SAs are provisioned within the BS. Dynamic SAs are created and destroyed in real time in response to the initiation and termination of specific service flows. Both Static and Dynamic SAs may be shared by multiple SSs. Each SS can have several service flows on the go and can therefore have multiple dynamic SAs. The BS always makes sure that the assigned SAs are compatible with the service types the SS is authorized to access. SA is identified by SAID, which contains Cryptographic suite (i.e., encryption algorithm) and Security Info (i.e., key, IV). The basic and primary management connections do not have SAs. The secondary management connection can have an optional SA. Transport connections always have SAs. An SA’s shared information includes the Cryptographic Suite employed within the SA. The shared information may include TEKs and Initialization Vectors. The exact content of the SA is dependent on the SA’s Cryptographic Suite. When SS establishes an exclusive Primary SA with its BS, the SAID for SS is equal to the Basic CID of that SS. Using the PKM protocol, an SS requests from its BS an SA’s keying material. An SA’s keying material (e.g., Data Encryption Standard (DES) key and CBC Initialization Vector) has a limited lifetime. When the BS delivers SA keying material to an SS, it also provides the SS with that material’s remaining lifetime.

228

R. Prasad and F.J. Velez

Fig. 5.9 PKMv1 protocol

Authorisation Protocol (AK Generation)

AK

KEK ( Derived from AK

Privacy & Key Management (TEK Generation )

TEKs

It is the responsibility of the SS to request new keying material from the BS before the set of keying material that the SS currently holds expires at the BS. In certain Cryptographic Suites, key lifetime may be limited by the exhaustion rate of a number space, for example, the Packet Number (PN) in AES-CCM mode. In this case, the key ends either at the expiry of the key lifetime or the exhaustion of the number space, whichever is earliest. Note that in this case, security is not determined by the key lifetime (Fig. 5.9).

5.5.1.2

SS Authorization and AK Exchange Overview

During the process of SS authorization the BS performs following steps: l l

l l

Authenticating a client SS’s identity Establishing a shared AK by RSA, from which a key encryption key (KEK) and message authentication keys are derived Providing the authenticated SS with the SAIDs Providing the SS with properties of primary and static SAs the SS is authorized to obtain keying information for

After achieving the initial authorization, SS needs to periodically reauthorize with the BS. The process of reauthorization is also managed by the SS’s Authorization state machine. TEK state machines manage the refreshing of TEKs. Authorization via RSA authentication protocol An SS begins authorization by sending an Authentication Information message to its BS. The Authentication Information message contains the SS manufacturer’s X.509 certificate, issued either by the manufacturer itself or by an external authority. The Authentication Information message is strictly informative, that is, the BS may choose to ignore it. However, it does provide a mechanism for a BS to learn the manufacturer certificates of its client SS.

5 Security Sublayer

229

SS

BS Authorisation Request (SS Certificate, Security Capabilities, SAID)

Verify SS Certificate

AK (128bits) Generation Authorisation Reply (AK, Key Lifetime, selected security suite, AK sequence number)

AK (128bits) Generation

AK (128bits) Generation Key lifetime = 1-70 days

Fig. 5.10 SS authorization using AK

The SS sends an Authorization Request message to its BS immediately after sending the Authentication Information message. This is a request for an AK, as well as for the SAIDs identifying any Static Security SAs the SS is authorized to participate in. The Authorization Request includes: 1. A manufacturer-issued X.509 certificate 2. A description of the cryptographic algorithms the requesting SS supports: SS’s cryptographic capabilities are presented to the BS as a list of cryptographic suite identifiers, each indicating a particular pairing of packet data encryption and packet data authentication algorithms the SS supports 3. The SS’s Basic CID: the Basic CID is the first static CID the BS assigns to an SS during initial ranging – the primary SAID is equal to the Basic CID In response to an Authorization Request message, a BS validates the requesting SS’s identity, determines the encryption algorithm and protocol support it shares with the SS, activates an AK for the SS, Fig. 5.10, encrypts it with the SS’s public key, and sends it back to the SS in an Authorization Reply message. The authorization reply includes the following components: l l

l l

An AK encrypted with the SS’s public key A 4-bit key sequence number, used to distinguish between successive generations of AKs A key lifetime The identities (i.e., the SAIDs) and properties of the single primary and zero or more static SAs the SS is authorized to obtain keying information for

230

R. Prasad and F.J. Velez

Authorization Reply identifies only Primary SA and Static SAs, not Dynamic ones. Before responding to the SS Authorization Request, the BS cross verifies whether the SS is authorized for basic unicast services and other provisioned services. This is done through X.509 certificate. AK is refreshed periodically by reissuing an Authorization Request to the BS. Reauthorization is identical to authorization with the exception that the SS does not send Authentication Information messages during reauthorization cycles. To avoid service interruptions during reauthorization, successive generations of the SS’s AKs have overlapping lifetimes. Therefore provision is made that both the SS and BS are capable of supporting up to two simultaneously active AKs during these transition periods. The protected services a BS makes available to a client SS can depend upon the particular cryptographic suites SS and BS share support for.

5.5.1.3

TEK Exchange

After the SS authentication procedure has been done, the AK is used to derive KEK and HMAC key. TEK is then generated by BS randomly. The TEK is the key actually used to encrypt data traffic exchanged between the BS and SS. A key exchange message is authenticated by HMAC-SHA1 to provide message integrity and AK confirmation. Two TEKs are used to encrypt data. The first one is used as the current operational key and the second one is used when the current TEK key expires.

TEK Exchange: for PtM Topology Upon achieving authorization, an SS starts a separate TEK state machine for each of the SAIDs (or per SA) identified in the Authorization Reply message. Each TEK state machine operating within the SS is responsible for managing the keying material associated with its respective SAID. TEK state machines periodically send Key Request messages to the BS, requesting a refresh of keying material for their respective SAIDs. The BS responds to a Key Request with a Key Reply message, containing the BS’s active keying material for a specific SAID, Fig. 5.11. The TEK is encrypted using appropriate KEK derived from the AK. Note that at all times the BS maintains two active sets of keying material per SAID. The lifetimes of the two generations overlap such that each generation becomes active halfway through the life of its predecessor and expires halfway through the life of its successor. A BS includes in its Key Replies both of an SAID’s active generations of keying material. The Key Reply provides the requesting SS, in addition to the TEK and CBC initialization vector (IV), the remaining lifetime of each of the two sets of keying material. The receiving SS uses these remaining lifetimes to estimate when the BS will invalidate a particular TEK, and therefore when to schedule future Key

5 Security Sublayer

231

SS

BS

AK (128bits)

AK (128bits)

KEK (128bits)

KEK (128bits) HMAC-Key(160bits)

HMAC-Key(160bits) TEK Key Request (AK Sequence Number, SAID, HMAC-SHA1)

TEK (128bits) Generation

TEK Key Reply (AK Sequence Number, SAID, HMAC-SHA1, Encrypted TEK, TEK key lifetime)

TEK TEK(128bits) (128bits)

TEK (128bits)

Key lifetime = 30mins to 7 days

Fig. 5.11 Data key exchange and TEK generation

Requests such that the SS requests and receives new keying material (before the BS expires the keying material the SS currently holds). The operation of the TEK state machine’s Key Request scheduling algorithm, combined with the BS’s regimen for updating and using an SAID’s keying material, ensures that the SS will be able to continually exchange encrypted traffic with the BS. A TEK state machine remains active as long as: l

l

The SS is authorized to operate in the BS’s security domain, that is, it has a valid AK The SS is authorized to participate in that particular SA, that is, the BS continues to provide fresh keying material during re key cycles

The parent Authorization state machine stops all of its child TEK state machines when the SS receives from the BS an Authorization Reject during a reauthorization cycle. Individual TEK state machines can be started or stopped during a reauthorization cycle if an SS’s Static SAID authorizations changed between successive re-authorizations. Communication between Authorization and TEK state machines occurs through the passing of events and protocol messaging. The Authorization state machine generates events (i.e., Stop, Authorized, Authorization Pending, and Authorization Complete ones) that are targeted at its child TEK state machines. TEK state machines do not target events at their parent Authorization state machine. The

232

R. Prasad and F.J. Velez

TEK state machine affects the Authorization state machine indirectly through the messaging a BS sends in response to an SS’s requests: a BS may respond to a TEK machine’s Key Requests with a failure response (i.e., Authorization Invalid message) to be handled by the Authorization state machine.

TEK Exchange: For Mesh Mode Upon achieving authorization, a Node starts for each Neighbour a separate TEK state machine for each of the SAIDs identified in the Authorization Reply message. Each TEK state machine operating within the Node is responsible for managing the keying material associated with its respective SAID. The Node is responsible for maintaining the TEKs between itself and all nodes it initiates TEK exchange with. Its TEK state machines periodically send Key Request messages to the Neighbours of the node, requesting a refresh of keying material for their respective SAIDs. The Neighbour replies to a Key Request with a Key Reply message, containing the BS’s active keying material for a specific SAID. The TEK in the Key Reply is encrypted, using the node’s public key found in the SS-Certificate attribute. The node maintains two active sets of keying material per SAID per neighbour at all times. The lifetimes of the two generations overlap such that each generation becomes active halfway through the life of its predecessor and expires halfway through the life of its successor. A neighbour includes in its Key Replies both of an SAID’s active generations of keying material. The Key Reply provides the requesting Node, in addition to the TEK, the remaining lifetime of each of the two sets of keying material. The receiving Node uses these remaining lifetimes to estimate when the Neighbour invalidates a particular TEK, and therefore when to schedule future Key Requests. The transmit regime between the initiating Node and the Neighbour one provides for seamless key transition.

State Machines for Key Exchange The request, generation and distribution of the encryption keys are complex processes that involve many states, messages, events, parameters and actions. State machine diagrams make this process simplified and systematic. Many different state machines are needed to describe the generation and distribution of the encryption keys. In the next section we describe the state machines for AK and TEK.

Authorization State Machine The Authorization state machine consists of six states and eight distinct events (including receipt of messages) that can trigger state transitions, Fig. 5.12. The

5 Security Sublayer

233 Timeout

Start

Auth Reject Wait

Silent

Auth Reject/

Communication Establised Auth Info,Auth Request

Perm Auth Reject/ Auth Reject / [TEK] stop

Auth Wait

Timeout/Auth Info, Auth Request Auth Reply/ [TEK] Authorised

Authorised

Perm Auth Reject / [TEK] stop

Reauth/ Auth Request

Auth Invalid / [TEK] Auth Pend

Auth Invalid/Auth Request, [TEK] Auth Pend Auth Grace Timeout/ Auth Request

Auth Replyo/ [TEK] Authorised,[TEK] Auth Comp,[TEK]Stop

Reauth Wait

Timeout /Auth Request

Fig. 5.12 Authorization state machine flow diagram

Authorization finite state machine (FSM) is presented below in a graphical format, as a state flow model and in a tabular format, as a state transition matrix. The state flow diagram depicts the protocol messages transmitted and internal events generated for each of the model’s state transitions; however, the diagram does not indicate additional internal actions, such as the clearing or starting of timers that accompany the specific state transitions. Accompanying the state transition matrix is a detailed description of the specific actions accompanying each state transition; the state transition matrix shall be used as the definitive specification of protocol actions associated with each state transition. The legends from Table 5.7 apply to the Authorization State Machine flow. The Authorization state transition matrix presented in Table 5.7 lists the six Authorization machine states in the topmost row and the eight Authorization machine events (includes message receipts) in the leftmost column. Any cell within the matrix represents a specific combination of state and event, with the next state (the state transitioned to) displayed within the cell. For example, cell 4-B represents the receipt of an Authorization Reply (Auth Reply) message when in the Authorize Wait (Auth Wait) state. Within cell 4-B is the name of the next state, “Authorized”. Thus, when an SS’s Authorization state machine is in the Auth Wait state and an Auth Reply message is received, the Authorization state machine will transition to the Authorized state. A shaded cell within the state transition matrix implies that either the specific event cannot or should not occur within that state. If the event does occur, the state machine should ignore it. For example, if an Auth Reply message arrives when in the Authorized state, that message should be ignored. The SS may, however, in response to an improper event, log its occurrence, generate an SNMP event, or take some other vendor-defined action. These actions,

234 Table 5.7 Authorization FSM state transition matrix State (A) (B) (C) Event or Rcvd Start Auth Wait Authorized Message (1) Communication Auth Wait Established (2) Auth Reject Auth Reject Wait (3) Perm Auth Silent Reject (4) Auth Reply Authorized (5) Timeout Auth Wait (6) Auth Grace Timeout (7) Auth Invalid (8) Reauth

R. Prasad and F.J. Velez

(D) (E) Reauth Wait Auth Reject Wait

(F) Silent

Auth Reject Wait Silent Authorized Reauth Wait Start Reauth Wait Reauth Wait Reauth Wait Reauth Wait

however, are not specified within the context of the Authorization state machine, which simply ignores improper events. The description of the FSM states, messages, events and parameters follows [12]: 1. States (a) Start – This is the initial state of the FSM. No resources are assigned to or used by the FSM in this state, for example, all timers are off, and no processing is scheduled. (b) Authorize Wait (Auth Wait) – The SS has received the “Communication Established” event indicating that it has completed basic capabilities negotiation with the BS. In response to receiving the event, the SS has sent both an Authentication Information and an Auth Request message to the BS and is waiting for the reply. (c) Authorized – The SS has received an Auth Reply message that contains a list of valid SAIDs for this SS. At this point, the SS has a valid AK and SAID list. Transition into this state triggers the creation of one TEK FSM for each of the SS’s privacy-enabled SAIDs. (d) Reauthorize Wait (Reauth Wait) – The SS has an outstanding reauthorization request. The SS was either about to expire (see Authorization Grace Time in Table 343) its current authorization or received an indication (an Authorization Invalid message from the BS) that its authorization is no longer valid. The SS sent an Auth Request message to the BS and is waiting for a response. (e) Authorize Reject Wait (Auth Reject Wait) – The SS received an Authorization Reject (Auth Reject) message in response to its last Auth Request. The Auth Reject’s error code indicated the error was not of a permanent nature. In response to receiving this reject message, the SS set a timer and transitioned to the Auth Reject Wait state. The SS remains in this state until the timer expires.

5 Security Sublayer

235

(f) Silent – The SS received an Auth Reject message in response to its last Auth Request. The Auth Reject’s error code indicated the error was of a permanent nature. This triggers a transition to the Silent state, where the SS is not permitted to pass subscriber traffic. The SS shall, however, respond to management messages from the BS issuing the Perm Auth Reject. 2. Messages (a) Authorization Request (Auth Request) – Request an AK and list of authorized SAIDs. It is sent from SS to BS. (b) Authorization Reply (Auth Reply) – Receive an AK and list of authorized, static SAIDs. Sent from BS to SS. The AK is encrypted with the SS’s public key; (c) Authorization Reject (Auth Reject) – Attempt to authorize was rejected. Sent from the BS to the SS. (d) Authorization Invalid (Auth Invalid) – The BS may send an Authorization Invalid message to a client SS as follows: l l

An unsolicited indication or A response to a message received from that SS

In either case, the Auth Invalid message instructs the receiving SS to re-authorize with its BS. The BS responds to a Key Request with an Auth Invalid message if (1) the BS does not recognize the SS as being authorized (i.e., no valid AK associated with SS) or (2) verification of the Key Request’s keyed message digest (in HMAC-Digest attribute) failed. Note that the Authorization Invalid event, referenced in both the state flow diagram and the state transition matrix, signifies either the receipt of an Auth Invalid message or an internally generated event. (e) Authentication Information (Auth Info) – The Auth Info message contains the SS manufacturer’s X.509 Certificate, issued by an external authority. The Auth Info message is strictly an informative message the SS sends to the BS; with it, a BS may dynamically learn the manufacturer certificate of client SS. Alternatively, a BS may require out-of-band configuration of its list of manufacturer certificates. 3. Events (a) Communication Established – The Authorization state machine generates this event upon entering the Start state if the MAC has completed basic capabilities negotiation. If the basic capabilities negotiation is not complete, the SS sends a Communication Established event to the Authorization FSM upon completing basic capabilities negotiation. The Communication Established event triggers the SS to begin the process of getting its AK and TEKs. (b) Timeout – A retransmission or wait timer timed out. Generally a request is resent. (c) Authorization Grace Timeout (Auth Grace Timeout) – The Authorization Grace timer timed out. This timer fires a configurable amount of time (the Authorization Grace Time) before the current authorization is supposed

236

R. Prasad and F.J. Velez

to expire, signalling the SS to reauthorize before its authorization actually expires. The Authorization Grace Time takes the default value or may be specified in a configuration setting within the Auth Reply message. (d) Reauthorize (Reauth) – SS’s set of authorized static SAIDs may have changed. This event is generated in response to an SNMP set and meant to trigger a reauthorization cycle. (e) Authorization Invalid (Auth Invalid) – This event is internally generated by the SS when there is a failure authenticating a Key Reply or Key Reject message, or externally generated by the receipt of an Auth Invalid message, sent from the BS to the SS. A BS responds to a Key Request with an Auth Invalid if verification of the request’s message authentication code fails. Both cases indicate BS and SS have lost AK synchronization. A BS may also send to an SS an unsolicited Auth Invalid message, forcing an Auth Invalid event. (f) Permanent Authorization Reject (Perm Auth Reject) – The SS receives an Auth Reject in response to an Auth Request. The error code in the Auth Reject indicates the error is of a permanent nature. What is interpreted as a permanent error is subject to administrative control within the BS. Auth Request processing errors that can be interpreted as permanent error conditions include the following: l

l l l

l

Unknown manufacturer (do not have CA certificate of the issuer of the SS Certificate) Invalid signature on SS certificate ASN.1 parsing failure Inconsistencies between data in the certificate and data in accompanying PKM data attributes Incompatible security capabilities

When an SS receives an Auth Reject indicating a permanent failure condition, the Authorization State machine moves into a Silent state, where the SS is not permitted to pass subscriber traffic. The SS shall, however, respond to management messages from the BS issuing the Perm Auth Reject. The SS shall also issue an SNMP Trap upon entering the Silent state. (g) Authorization Reject (Auth Reject) – The SS receives an Auth Reject in response to an Auth Request. The error code in the Auth Reject does not indicate the failure was due to a permanent error condition. As a result, the SS’s Authorization state machine shall set a wait timer and transition into the Auth Reject Wait State. The SS shall remain in this state until the timer expires, at which time it shall reattempt authorization. (h) Authorization Complete (Auth Comp) – Sent by the Authorization FSM to a TEK FSM in the Operational Reauthorize Wait (Op Reauth Wait) or Rekey Reauthorize Wait (Rekey Reauth Wait) states to clear the wait state begun by a TEK FSM Authorization Pending event.

5 Security Sublayer

237

4. Parameters All configuration parameter values take the default values from Table 5.7 or may be specified in the Auth Reply message. (a) Authorize Wait Timeout (Auth Wait Timeout) – Timeout period between sending Authorization Request messages from Auth Wait state. (b) Authorization Grace Timeout (Auth Grace Timeout) – Amount of time before authorization is scheduled to expire that the SS starts reauthorization. (c) Authorize Reject Wait Timeout (Auth Reject Wait Timeout) – Amount of time an SS’s Authorization FSM remains in the Auth Reject Wait state before transitioning to the Start state.

5.5.1.4

TEK State Machine

The TEK state machine consists of six states and nine events (including receipt of messages) that can trigger state transitions. Like the Authorization state machine, the TEK state machine is presented in both a state flow diagram and a state transition matrix. As was the case for the Authorization state machine, the state transition matrix shall be used as the definitive specification of protocol actions associated with each state transition. Shaded states (Operational, Rekey Wait, and Rekey Reauthorize Wait) have valid keying material and encrypted traffic can be passed. The Authorization state machine starts an independent TEK state machine for each of its authorized SAIDs. The BS maintains two active TEKs per SAID. The BS includes in its Key Replies both of these TEKs, along with their remaining lifetimes. The BS encrypts downlink traffic with the older of its two TEKs and decrypts uplink traffic with either the older or newer TEK, depending upon which of the two keys the SS was using at the time. The SS encrypts uplink traffic with the newer of its two TEKs and decrypts downlink traffic with either the older or newer TEK, depending upon which of the two keys the BS was using at the time. Through operation of a TEK state machine, the SS attempts to keep its copies of an SAID’s TEKs synchronized with those of its BS. A TEK state machine issues Key Requests to refresh copies of its SAID’s keying material soon after the scheduled expiration time of the older of its two TEKs and before the expiration of its newer TEK. To accommodate for SS/BS clock skew and other system processing and transmission delays, the SS schedules its Key Requests a configurable number of seconds before the newer TEK’s estimated expiration in the BS. With the receipt of the Key Reply, the SS shall always update its records with the TEK Parameters from both TEKs contained in the Key Reply message. The SS’s scheduling of its key refreshes in conjunction with its management of an SA’s active TEKs. The six states in TEK state machine are Start, Op Wait, Op Reauth Wait, Operational, Rekey Wait, and Rekey Reauth Wait, Table 5.8. Their description follows:

238

R. Prasad and F.J. Velez

Table 5.8 Table Authorization FSM state transition matrix State Event or (A) (B) (C) (D) Rcvd Message Start Op Wait Op Reauth Op Wait (1) Stop (2) Authorized Op Wait (3) Auth Pend

(4) Auth Comp (5) TEK Invalid (6) Timeout (7) TEK Refresh Timeout (8) Key Reply (9) Key Reject

Start

Start

Start

Op Reauth Wait

(E) Rekey Wait Start

(F) Rekey Reauth Wait Start

Rekey Reauth Wait Op Wait Op Wait

Op Wait

Op Wait

Rekey Wait Op Reauth Wait

Rekey Wait Rekey Wait

Operational Start

Operational Start

1. Start – This is the initial state of the FSM. No resources are assigned to or used by the FSM in this state – for example, all timers are off, and no processing is scheduled. 2. Operational Wait (Op Wait) – The TEK state machine has sent its initial request (Key Request) for its SAID’s keying material (TEK and CBC initialization vector), and is waiting for a reply from the BS. 3. Operational Reauthorize Wait (Op Reauth Wait) – The wait state the TEK state machine is placed in if it does not have valid keying material while the Authorization state machine is in the middle of a reauthorization cycle. 4. Operational – The SS has valid keying material for the associated SAID. 5. Rekey Wait – The TEK Refresh Timer has expired and the SS has requested a key update for this SAID. Note that the newer of its two TEKs has not expired and can still be used for both encrypting and decrypting data traffic. 6. Rekey Reauthorize Wait (Rekey Reauth Wait) – The wait state the TEK state machine is placed in if the TEK state machine has valid traffic keying material, has an outstanding request for the latest keying material, and the Authorization state machine initiates a reauthorization cycle.

5.5.1.5

TEK Exchange Overview for PtM Topology

If the SS and BS decide “No authorization”, as their authorization policy, the SS and BS shall perform neither SA-TEK handshake nor Key Request/Key Reply

5 Security Sublayer

239

handshake. In this case, target SAID value, which may be included in DSA-REQ/ RSP messages, shall be Null SAID. Upon achieving authorization, an SS starts a separate TEK state machine for each of the SAIDs identified in the Authorization Reply or PKMv2 SA-TEK-RSP message, if data traffic encryption is provisioned for one or more service flows. Each TEK state machine operating within the SS is responsible for managing the keying material associated with its respective SAID. TEK state machines periodically send Key Request messages to the BS, requesting a refresh of keying material for their respective SAIDs. The BS responds to a Key Request with a Key Reply message, containing the BS’s active keying material for a specific SAID. TEKs and KEKs may be either 64 bits or 128 bits long. SAs employing any ciphersuite with a basic block size of 128 bits shall use 128-bit TEKs and KEKs. Otherwise 64-bit TEKs and KEKs shall be used. The name TEK-64 is used to denote a 64-bit TEK and TEK-128 is used to denote a 128-bit TEK. Similarly, KEK-64 is used to denote a 64-bit KEK and KEK-128 is used to denote a 128-bit KEK. For SAs using a ciphersuite employing DES-CBC, the TEK in the Key Reply is triple DES (3-DES) (encrypt-decrypt-encrypt or EDE mode) encrypted, using a two-key, 3-DES KEK derived from the AK. For SAs using a ciphersuite employing 128 bits keys, such as AES-CCM mode, the TEK in the key Reply is AES encrypted using a 128-bit key derived from the AK and a 128-bit block size. Note that at all times the BS maintains two diversity sets of keying material per SAID. The lifetimes of the two generations overlap such that each generation becomes active halfway through the life of it predecessor and expires halfway through the life of its successor. A BS includes in its Key Replies both of an SAID’s active generations of keying material. For SAs using a ciphersuite employing CBC mode encryption the Key Reply provides the requesting SS, in addition to the TEK and CBC initialization vector, the remaining lifetime of each of the two sets of keying material. For SAs using a ciphersuite employing AES-CCM mode, the Key Reply provides the requesting SS, in addition to the TEK, the remaining lifetime of each of the two sets of keying material. The receiving SS uses these remaining lifetimes to estimate when the BS will invalidate a particular TEK, and therefore when to schedule future Key Requests such that the SS requests and receives new keying material (before the BS expires the keying material the SS currently holds). For AES-CCM mode, when more than half the available PN numbers in the 31-bit PN number space are exhausted, the SS shall schedule a future Key Request in the same fashion as if the key lifetime was approaching expiry. The operation of the TEK state machine’s Key Request scheduling algorithm, combined with the BS’s regimen for updating and using an SAID’s keying materia ensures that the SS will be able to continually exchange encrypted traffic with the BS (Table 5.9).

240

R. Prasad and F.J. Velez

Table 5.9 Different keys used in IEEE 802.16 Key Generated Used for by Authentication BS Generating KEKs Calculating Key (AK) HMAC digests Verifying received HMAC digests Key Encryption BS, SS Encrypting TEK for Key (KEK) transmission (BS) Decrypting TEK for use (SS) Traffic BS Encrypting data traffic Encryption Key (TEK)

5.6

Lifetime

Algorithm

1–70 days

3-DES SHA-1

Same as AK

3-DES

30 min to 7 days

DES CBC

Key Management

Key management is handled by the authorisation and TEK state machines in the SS. The authorization SA has a 60-bit authorization key (AK) and uses a 4-bit quantity to identify the AK. It uses an X.509 certificate to identify SS. The default lifetime of AK is 7 days and it could be set from 1 to 70 days.

5.6.1

Authorisation Key (AK) Management

The BS is responsible for maintaining keying information for all SAs. This section describes a mechanism for synchronizing this keying information between the BS and its client SS. After an SS completes basic capabilities negotiation with the BS, it sends an Authorisation Request (Auth Request) message. When the BS receives an unauthorised SS request it initiates the activation of a new AK. This is send through the Authorisation Reply message. This AK remains active until it expires according to its predefined AK Lifetime which is a BS system configuration parameter. In case the SS fails to reauthorize before the expiration of its current AK, the BS does not hold active AKs for the SS and SS is declared as unauthorized. The BS therefore removes all associated TEKs from its keying tables for such an unauthorized SS. Using AK transition period the BS is always prepared to send an AK to an SS upon request. The SS is responsible for requesting authorisation with its BS and maintaining an active AK. An SS refreshes its AK by reissuing an Authorisation Request to the BS. Two simultaneous AKs for different clients are handled by the BS. The BS has two active AKs during an AK transition period; the two active keys have overlapping lifetimes. An AK transition period begins when the BS receives an Auth

5 Security Sublayer

241

Request message from an SS and the BS has a single active AK for that SS as shown in the figure. In response to this Auth Request, the BS activates a second AK. This AK has a key sequence number one larger than that of the existing AK and is sent back to the requesting SS along with Auth Reply message. The active lifetime of the second AK is set by the BS as the remaining lifetime of the first AK plus the predefined AK Lifetime. The second AK therefore remains active for one AK Lifetime beyond the expiration of the first, “older” key. The key transition period ends with the expiration of the older key. As long as the BS is holding two active AKs and is in the midst of an SS’s AK transition period, it responses to Auth Request messages with the newer of the two active keys. When the older key expires, Auth Request triggers the activation of a new AK, and the start of a new key transition period.

5.6.1.1

BS Usage of AK

The BS uses keying material derived from the SS’s AK for the following: l l

l

Verifying the HMAC-Digests in Key Request messages received from that SS Calculating the HMAC-Digests it writes into Key Reply, Key Reject, and TEK Invalid messages sent to that SS Encrypting the TEK in the Key Reply messages it sends to that SS

HMAC_KEY_U derived from one of the SS’s active AKs is used to verify the HMAC-Digest in Key Request messages received from the SS. The AK Key Sequence Number accompanying each Key Request message allows the BS to determine which HMAC_KEY_U was used to authenticate the message. If the AK Key Sequence Number indicates the newer of the two AKs, the BS shall identify this as an implicit acknowledgment that the SS has obtained the newer of the SS’s two active AKs. BS shall use an HMAC_KEY_D derived from the active AK selected above when calculating HMAC-Digests in Key Reply, Key Reject, and TEK Invalid message. When sending Key Reply, Key Reject, or TEK Invalid messages within a key transition period, if the newer key has been implicitly acknowledged, the BS shall use the newer of the two active AKs. If the newer key has not been implicitly acknowledged, the BS shall use the older of the two active AKs to derive the KEK and the HMAC_KEY_D. The BS shall use a KEK derived from an active AK when encrypting the TEK in the Key Reply messages. The right-hand side of Fig. 5.13 shows the BS’s policy regarding its use of AKs, where the shaded portion of an AK’s lifetime indicates the time period during which that AK shall be used to derive the HMAC_KEY_U, HMAC_KEY_D, and KEK. For calculating the HMAC-Digest in the HMAC Tuple attribute, the BS shall use the HMAC_KEY_U and HMAC_KEY_D derived from one of the active AKs. For signing messages, if the newer AK has been implicitly acknowledged, the BS shall

242

R. Prasad and F.J. Velez SS

BS Authenticat

ion Informat ion Authorizatio n Request

K 0}

n Reply {A

Authorizatio

Authorizatio

x

n Request

AK0 Active Lifetime AK Grace time

AK0 Lifetime

K 1}

n Reply {A

Authorizatio

a

Key Reque

st {AK } 1 b c

y AK1 Active Lifetime

Authorizatio n Request Authorizatio n (re)Reque st

AK1 Lifetime d

K 2}

n Reply {A Authorizatio

AK Grace time

e

Key Reque

st {AK } 2 f AK2 Lifetime

AK2 Active Lifetime

switch over point AK used to en/decrypt TEK

Fig. 5.13 AK management in BS and SS (adapted from [3])

use the newer of the two active AKs to derive the HMAC_KEY_D. If the newer key has not been implicitly acknowledged, the BS shall use the older of the two active AKs to derive the HMAC_KEY_D. The HMAC Key Sequence Number in the HMAC Tuple, equal to the AK’s sequence number from which the HMAC_KEY_D was derived, enables the SS to correctly determine which HMAC_KEY_D was used for message authentication.

5 Security Sublayer

243

When receiving messages containing the HMAC Tuple attribute, the BS shall use the HMAC_KEY_U indicated by the HMAC Key Sequence Number to authenticate the messages.

5.6.1.2

TEK Lifetime

The BS maintains two sets of active TEKs (and their associated Initialization Vectors, or IVs) per SAID, corresponding to two successive generations of keying material. The two generations of TEKs shall have overlapping lifetimes determined by TEK Lifetime, a predefined BS system configuration parameter. The newer TEK shall have a key sequence number one greater (modulo 4) than that of the older TEK. Each TEK becomes active halfway through the lifetime of its predecessor and expires halfway through the lifetime of its successor. Once a TEK’s lifetime expires, the TEK becomes inactive and shall no longer be used. The Key Reply messages sent by a BS contain TEK parameters for the two active TEKs. The TEKs’ active lifetimes that a BS reports in a Key Reply message shall reflect, as accurately as an implementation permits, the remaining lifetimes of these TEKs at the time the Key Reply message is sent. 5.6.1.3

BS Usage of TEK

The BS transitions between the two active TEKs differently, depending on whether the TEK is used for downlink or uplink traffic. For each of its SAIDs, the BS shall transition between active TEKs according to the following rules: l

l

At expiration of the older TEK, the BS shall immediately transition to using the newer TEK for encryption. The uplink transition period begins from the time the BS sends the newer TEK in a Key Reply message and concludes once the older TEK expires.

It is the responsibility of the SS to update its keys in a timely fashion; the BS shall transition to a new downlink encryption key regardless of whether a client SS has retrieved a copy of that TEK. The BS uses the two active TEKs differently, depending on whether the TEK is used for downlink or uplink traffic. For each of its SAIDs, the BS shall use the two active TEKs according to the following rules: l

l

The BS shall use the older of the two active TEKs for encrypting downlink traffic. The BS shall be able to decrypt uplink traffic using either the older or newer TEK.

Note that the BS encrypts with a given TEK for only the second half of that TEK’s total lifetime. The BS is able, however, to decrypt with a TEK for the TEK’s entire lifetime.

244

5.6.1.4

R. Prasad and F.J. Velez

Node Re-Authorization in Mesh Mode During Normal Operation

When re-authorizing with the network, the re-authorizing node shall tunnel the authorization messages.

5.6.2

SS Key Usage

For sustaining authorisation the SS is responsible for maintaining an active AK with its BS. An SS uses its two most recently obtained AKs according to the following manner.

5.6.2.1

SS Reauthorization

AKs are refreshed regularly as they have a limited lifetime. An SS refreshes its AK by reissuing an Auth Request to the BS. The Authorization State Machine manages the scheduling of Auth Requests for refreshing AKs. An SS’s Authorization state machine schedules the beginning of reauthorization a configurable duration of time, the Authorization Grace Time, before the SS’s latest AK is scheduled to expire. The Authorization Grace Time is configured to provide an SS with an authorization retry period that is sufficiently long to allow for system delays and provide adequate time for the SS to successfully complete an Authorization exchange before the expiration of its most current AK.

5.6.2.2

SS Usage of AK

An SS shall use the HMAC_KEY_U derived from the newer of its two most recent AKs when calculating the HMAC-Digests it attaches to Key Request messages. The SS shall be able to use the HMAC_KEY_D derived from either of its two most recent AKs to authenticate Key Reply, Key Reject, and TEK Reject messages. The SS shall be able to decrypt an encrypted TEK in a Key Reply message with the KEK derived from either of its two most recent AKs. The SS shall use the accompanying AK Key Sequence Number to determine which set of keying material to use. The left-hand side of Fig. 5.14 illustrates an SS’s maintenance and usage of its AKs, where the shaded portion of an AK’s lifetime indicates the time period during which that AK shall be used to decrypt TEKs. Even though it is not part of the message exchange, Fig. 5.14 also shows the implicit acknowledgment of the reception of a new AK via the transmission of a Key Request message using the key sequence of the new AK. An SS shall use the HMAC_KEY_U derived from the newer of its two most recent AKs when calculating the HMAC-Digests of the HMAC Tuple attribute.

5 Security Sublayer

245

SS

BS

Key Request

EK0,TEK1}

Key Reply {T

TEK0 Active Lifetime

TEK0 Active Lifetime TEK1 Active Lifetime

TEK1 Active Lifetime

Key Reques

t

TEK Grace Time

EK0,TEK1}

Key Reply {T

TEK2 Active Lifetime

TEK2 Active Lifetime

TEK Grace time

TEK3 Active TEK Lifetime Grace time TEK0 Active Lifetime

Key Req

uest

Key (re)

Request TEK3 Active Lifetime

EK0,TEK1}

Key Reply {T

Key Request Key Reply {TEK0,TEK1}

TEK0 Active Lifetime TEK1 Active Lifetime

Switch over point TEK used for encryption

Fig. 5.14 TEK management in BS and SS (adapted from [3])

5.6.2.3

SS Usage of TEK

An SS shall be capable of maintaining two successive sets of traffic keying material per authorized SAID. Through operation of its TEK state machines, an SS shall request a new set of traffic keying material a configurable amount of time, the TEK Grace Time [see points (x) and (y) in Figure 134], before the SS’s latest TEK is scheduled to expire.

246

R. Prasad and F.J. Velez

For each of its authorized SAIDs, the SS: l l

Shall use the newer of its two TEKs to encrypt uplink traffic Shall be able to decrypt downlink traffic encrypted with either of the TEKs

The left-hand side of Fig. 5.14 illustrates the SS’s maintenance and usage of an SA’s TEKs, where the shaded portion of a TEK’s lifetime indicates the time period during which that TEK shall be used to encrypt MAC PDU payloads.

5.6.3

PKM Version 2

The PKMv1 supports only the device authentication and had many critical drawbacks. Therefore IEEE 802.16e released the second version of Privacy and Key Management (PKM) protocol called PKMv2. PKMv2 aims to fix the bugs in the former version PKMv1.The new released version of the PKM protocol provides mechanisms for the double user/device authentication. It has many security features like message authentication codes, key ids, certificates, etc. The IEEE 802.16e-2005 standard supports PKMv2 which allows for three types of authentication: l l l

RSA based authentication EAP based authentication (optional) RSA based authentication followed by EAP authentication

Compared to PKMV1 which only supported RSA for Authentication/Authorisation, PKMv2 supports EAP also. The PKMv2 comprises of two main parts, an Authentication/Authorization protocol to establish a shared Authorization Key (AK), and a 3-Way Security Association (SA) Traffic Encryption Key (TEK) Handshake. As shown in the Fig. 5.15, the first part of the PKMv2 protocol an Authorization Key (AK) is generated using RSA or EAP or both. After that, each party (SS as well as BS generates a Key Encryption Key (KEK) using their AKs. KEKs are used in AK Generation is established using Either EAP or RSA or both

AK also KEKs and H-C/MAC keys are derived from AK

PKMv2 SA-TEK 3-Way Handshake

TEKs

Fig. 5.15 PKMv2 protocol

5 Security Sublayer

247

encrypting and distributing Traffic Encryption Keys (TEK), TEKs can be taken as session keys, while AK/KEK are long term keys. The important part of PKMv2 is the SA-TEK 3-Way Handshake. Through SATEK 3-Way Handshake SS (here MS, used interchangeably in the entire text) and BS exchange TEKs. In the handshake, TEKs are encrypted by KEKs. The EAP protocol can support a large set of methods, such as EAP-TLS, EAP-AKA and EAP-TTLS. PKMv2 uses a slightly different key hierarchy. This is due to the fact that two authentication systems are used one based on RSA and the other on EAP. TEKs are derived in the same way as in PKMv1 except that their names are different. The RSA-generated key is called pre-primary AK (pre-PAK) and the one generated through EAP based authentication is called Master Session Key(MSK) The general procedure to derive Authentication Key (AK) depends on authentication scheme used as listed below: l

If RSA-based authorization then AK = Dot16KDF (PAK, SS MAC address|BSID|PAK|“AK”,160)

l

If EAP-based authorization then AK = Dot16KDF (PMK, SS MAC address|BSID|“AK”, 160)

l

If RSA-based and EAP-based authorization then, AK=Dot16KDF (PAK xor PMK, SS MAC address|BSID|PAK|“AK”,160)

l

If EAP-in-EAP then, AK = Dot16KDF (PMK xor PMK2, SS MAC address | BSID | “AK”, 160) Table 5.10 presents a summary of the comparisons between PKMv1 and PKMv2.

5.7

Network Aspects of Security

Access control mechanism ensures that only valid users are allowed access to the network. Figure 5.16 shows a typical access control architecture. Table 5.10 Comparison of PKMv1 and PKMv2 PKMv1 Authentication direction Unilateral Authentication method RSA based TEK encryption 3DES, AES, RSA Authentication object Keys involved in authorization Data encryption Data integrity

MS/SS AK No encryption, DES, AES-CCM No MAC, HMAC

PKMv2 Bilateral RSA based, EAP based 3DES, AES, RSA AES-ECB/ KEY-WRAP MS/SS, BS RSA based: Pre-PAK,AK EAP based: MSK, PMK, AK No encryption, DES, 3DES, AES-CCM/CBC/CTR No MAC, HMAC/CMAC

248

R. Prasad and F.J. Velez EAP

EAP

WiMAX Link

AAA-RADIUS/DIAMETER

user1

user2

Authenticator (Network Access Server)

IP cloud

Authentication Server

user3

Fig. 5.16 Access control architecture and authentication

The authentication is based on a three-party model: the supplicant, which requires access; the authenticator, which grants access; and the authentication server, which gives permission. Thus the three elements make an access control mechanism: l

l l

User/Client/Supplicant – The party in the authentication process that will provide its identity, and evidence for it and, as a result, will be authenticated. This party may also be referred to as the authenticating user, or the client. Authenticator – is the party that controls the access gate. Authentication server – is the entity that validates client credentials and determines if the client should be allowed access. The authenticator itself does not know whether an entity can be allowed access, it’s the authentication server who decides it.

5.7.1

Radius

Remote Authentication Dial-In User Service (RADIUS) is widely used standard for communication between the authenticator and authentication server. It has client/ server architecture and utilizes UDP messages .The authentication server is also the RADIUS server whereas the authenticator acts as RADIUS client. RADIUS is an IETF standard [17, 18]. In addition to authentication, RADIUS also supports authorization and accounting functions [18]. The authentication, authorization and accounting functions are collectively referred to as AAA functions. IETF has also developed a new standard for AAA functions named as DIAMETER [19] which tries to overcome the deficiencies in RADIUS. DIAMETER has higher reliability, security, and roaming support compared to RADIUS [18].

5.7.2

Extensible Authentication Protocol (EAP)

EAP is flexible framework defined by IETF (RFC 3748) which allows complex authentication protocols to be exchanged between the authenticator and the end user [16], Fig. 5.16.

5 Security Sublayer

249

EAP was introduced which can offer an authentication scheme to prevent the above mentioned problems. It integrates different authentication methods to match the nature of the communication channel. These methods are advised by IEEE including EAP-PKM, EAP-MD5, EAP-OTP, EAP-GTC, EAP-TLS, EAP-SIM and EAP-AKA. In mobile WiMAX, authentication is achieved using one of several possible EAP methods. Table 5.11 presents a list of the Wireless network security concerns and their solutions for WIMAX. Table 5.12 presents the difference between the security aspects in IEEE 802.16d and 802.16e. Table 5.11 Wireless network security concerns and their solutions for WIMAX Concern for Security Concern Purpose Implementation of the stakeholder respective security concern in IEEE 802.16 Network User Privacy Protect from interception RSA encryption, EAP-TLS, concern and eavesdropping PKM protocol Message Provides integrity of the X.509 Authentication message and sender authentication Data Integrity Protect user data from RSA encryption, EAP-TLS, being tampered in PKM protocol transit Access to services Allow only the user with X.509, EAP the right credentials service access Correct accounting Accuracy and efficiency AAA architecture of accounting Network User authentication Is the user who he says X.509, EAP-TTLS Operator he is? Concerns Device Is the device the correct X.509, EAP-TTLS authentication device? Authorization Is the user authorized to RSA, EAP, PKMv2 receive a particular protocol service? Access control Is the user authorized to RSA, EAP, PKMv2 receive this service? protocol Correct accounting Accuracy and efficiency AAA architecture of accounting

Table 5.12 Difference between IEEE 802.16d and 802.16e IEEE 802.16d PKM version PKM v1 Supported Mode Unicast Authorization method X.509 cert Authentication Unilateral Security algorithms RSA/DES/3DES/AES

IEEE 802.16e PKM v1/PKM v2 Unicast/Multicast/Broadcast X.509 cert/EAP Mutual/Unilateral RSA/DES/3DES/AES

250

5.8

R. Prasad and F.J. Velez

Conclusions

This Chapter starts by addressing aspects of the security architecture for IEEE 802.16, including its functions, secure encapsulation of MPDUs, encryption algorithms, data encryption methodologies (with DES in CBC mode, AES in CCM mode and the ones for IEEE 802.11e). Then, the authentication protocol (RSA and PKM EAP) was addressed and the PKM protocol presents. After describing the PKM protocol MAC management messages, addresses key management was discussed. The network aspects of security were finally presented, including RADIUS and Extensible Authentication Protocol (EAP).

References 1. C. Eklund, R.B. Marks, K.L. Stanwood, S. Wang, IEEE Standard 802.16: A Technical Overview of the WirelessMAN Air Interface for Broadband Wireless Access (June 2002) 2. D. Johnston, J. Walker, Overview of IEEE 802.16 security. IEEE Secur. Privacy Mag. 2(3), 40–48 (2004) 3. IEEE Std 802.16e-2005, 2006. Standard for Local and metropolitan area networks Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems Amendment 2: Physical and Medium Access Control Layers for Combined Fixed and Mobile Operation in Licensed Bands and Corrigendum 1, IEEE, New York, USA 4. K. Pahlavan, P. Krishnamurthy, Principles of Wireless Networks: A unified Approach (Pearson Education, Prentice Hall PTR, Upper Saddle River, NJ, 2002) 5. W. Stalling, Cryptography and Network Security: Principles and Practices, 3rd edn. (Pearson Education, Prentice Hall PTR, Upper Saddle River, NJ, 2003) 6. PKCS#1 v2.0,RSA Cryptography Standard, RSA Laboratories (Oct 1998) 7. National Technical Information Service (NTIS), FIPS 46-3, Data Encryption Standard (DES) (Oct 1999). http://www.ntis.gov/ 8. National Technical Information Service (NTIS), FIPS 197, Advanced Encryption Standard (AES) (Nov 2001) 9. National Technical Information Service (NTIS), FIPS 198, The key Hash Message Authentication Code (HMAC) (Mar 2002) 10. H. Krawczyk, M. Bellare, R. Canetti, IETF RFC. HMAC: Keyed-Hashing for Message Authentication (Feb 1997) 11. J.H. Song et al., IETF RFC 4493. The AES-CMAC Algorithm (June 2006) 12. IEEE 802.16 Broadband Wireless Access Working Group. Mapping of authentication Rejection from PKMv1 to PKMv2, IEEE (April 2008). http://www.ieee802.org/16/maint/ 13. FIPS 74, Guidelines for Implementing and Using the NBS Data Encryption Standard (Apr 1981) 14. FIPS 81, DES Modes of Operation (Dec 1980) 15. IETF RFC 3610, Counter with CBC-MAC (CCM) (D. Whiting et al., Sept 2003) 16. M. Barbeau, WiMax/802.16 Threat Analysis, School of Computer Science, Carleton University, Ottawa, Canada (October 2005) 17. C. Rigney et al., Remote dial-in user service (RADIUS). IETF RFC 2865 (June 2000) 18. A. Ghost, R. Muhamed, J.G. Andrews, Fundamentals of WiMAX: Understanding Broadband Wireless Networking (Prentice Hall, Upper Saddle River, NJ, 2007) 19. P. Calhoun et al., Diameter in use. IETF RFC 3588 (Sept 2003) 20. S. Xu, M. Matthews, C.-T. Huang, Security issues in privacy and key management protocols of IEEE 802.16, in Proceedings of the 44th ACM Southeast Conference (ACMSE 2006) (Mar 2006)

Chapter 6

Mobility Management Architecture for WiMAX Networks Susana Sargento, Pedro Neves, Ricardo Matos, Marı´lia Curado, Bruno Sousa, Kostas Pentikousis, and Giada Landi

Abstract Although WiMAX, based on the IEEE 802.16 family of standards, has emerged as one of the major candidates for next generation networks, it is also clear that in the near future, the combination of several technologies will be required. In this sense, the support of mobility in heterogeneous environments, addressing broadband wireless, is one of the main requirements in next generation networks. This chapter presents an architecture based on the recently standardized IEEE 802.21 framework, integrating both mobility and Quality of Service (QoS) mechanisms, and accommodating different wired and wireless technologies, such as WiMAX, Wi-Fi, DVB, and UMTS. This architecture supports seamless mobility in broadband wireless access (BWA) networks, and thus, it is suitable for next generation network environments. The results, obtained through real experimentation of the implemented architecture through an advanced mobility scenario using a real WiMAX testbed, show that the architecture is able to provide QoS under dynamic scenarios, with fast integrated QoS and mobility signaling.

Abbreviations AAA AC AF AIP AP ASN ASN-GW

Authentication, Authorization and Accounting Admission Control Application Function All-IP Access Point Access Service Network ASN-Gateway

S. Sargento (*) Institute of Telecommunications, University of Aveiro, Campus Universita´rio de Santiago, 3810193 Aveiro, Portugal e-mail: [email protected]

R. Prasad and F.J. Velez, WiMAX Networks, DOI 10.1007/978-90-481-8752-2_6, # Springer ScienceþBusiness Media B.V. 2010

251

252

BE BS BWA CID COTS CPS CS CSC CSN DHCP DNS ertPS FA FBSS FMIP GA GIST GLSM HA HO HHO ID IMS IP LLC LSIM PHY MAC MDHO MICS MIES MIH MIHF MIHU MIIS MIP MM MN MRI MS NCMS NGN NRM nrtPS

S. Sargento et al.

Best Effort Base Station Broadband Wireless Access Connection Identifier Commercial-of-the-Shelf Common Part Sublayer Convergence Sublayer Connectivity Service Controller Connectivity Service Network Dynamic Host Configuration Protocol Domain Name Service Extended real time Polling Service Foreign Agent Fast Base Station Switching Fast Handovers for MIP Generic Adapter General Internet Signaling Transport Generic Link State Monitor Home Agent Handover Hard Handover IDentifier IP Multimedia Subsystem Internet Protocol Link Lower Controller Link-specific Information Monitor Physical Medium Access Control Macro Diversity Handover Media Independent Command Service Media Independent Event Service Media Independent Handover Media Independent Handover Function Media Independent Handover User Media Independent Information Service Mobile IP Mobility Management Mobile Node Message Routing Information Mobile Station Network Control and Management System Next Generation Network Network Reference Model Non-real-time Polling Service

6 Mobility Management Architecture for WiMAX Networks

NSIS NSLP NWG PMIP PoA PoS QNE QoS QSPEC RC RMF RP RSA rtPS SAP SDP SDU SF SFM SHO SIP SNMP SS TCP TFPT UDP UGS WA WEIRD WiMAX

6.1

253

Next Steps in Signaling NSIS Signaling Layer Protocol Network Working Group Proxy MIP Point of Attachment Point of Service QoS Node Entity Quality of Service QoS SPECification Resource Controller Resource Management Function Reference Point Redline Specific Adapter real time Polling Service Service Access Point Session Description Protocol Service Data Units Service Flow Service Flow Management Soft Handover Session Initiation Protocol Simple Network Management Protocol Subscriber Station Transmission Control Protocol Trivial File Transfer Protocol User Datagram Protocol Unsolicited Grant Service WEIRD Agent WiMAX Extension to Isolated Research Data networks Worldwide Interoperability for Microwave Access

Introduction

Broadband Wireless Access (BWA) technologies are expected to play a central role in Next Generation Networks (NGN) [1, 2]. WiMAX [3], based on the IEEE 802.16 family of standards [4, 5], is one such technology that has the potential to form the foundation upon which operators will deliver ubiquitous Internet access in the near future, greatly benefiting from existing and future infrastructures. One of the central concerns in the emerging telecommunications environment is the support for seamless mobility while taking advantage of different access networks, some of which are already in use, others, such as WiMAX, will soon be deployed.

254

S. Sargento et al.

There are several proposals for fast and seamless mobility management between different access networks. IEEE has been working on the 802.21 draft standard [6] to enable Media Independent Handovers (MIH). IEEE 802.21 defines an abstract framework which delivers link layer information to the higher layers, in an effort to optimize handovers between heterogeneous media. We could argue that IEEE 802.21 aspires to harmonize mobility management processes, irrespective of the underlying technologies, considering that proper communication and interfaces are presented to the link layers. Although the work within the IEEE 802.21 working group is already in an advanced stage, the framework needs to be integrated with specific technologies, since each one has its particular mobility control procedures. Moreover, seamless mobility requires the active support of Quality of Service (QoS) related mechanisms in the handover process, guaranteeing that resources are reserved in the target access network before mobility management operations are completed. In other words, we cannot dissociate mobility management and QoS processes. This chapter presents mobility architecture with integrated QoS support, based on IEEE 802.21. The proposed architecture can accommodate different wired and wireless technologies, such as WiMAX, Wi-Fi, DVB, and UMTS. The way mobility management processes are intertwined with QoS mechanisms in order to support seamless mobility over heterogeneous networks is described. More specifically, we explain how QoS mechanisms based on the Next Steps in Signaling (NSIS) protocol family [7–9] can be integrated with IEEE 802.21 concepts. Finally, a prototype implementation of the proposed architecture is presented and results from its empirical evaluation on a state-of-the-art testbed using commercial off-the-shelf (COTS) WiMAX equipment are provided. By using a real demonstrator, with a WiMAX backhaul, we measured the processing time for each module involved in the handover process while discussing the obtained results. This chapter is organized as follows. Section 6.2 reviews the current state of the art in WIMAX mobility management architectures, reporting on recent advances in standardization for a, such as IEEE, WiMAX Forum and IETF. Section 6.3 introduces the proposed mobility-QoS integrated architecture, its elements and functionalities, taking as a use case the architecture developed within the WEIRD project [10]. Section 6.4 briefly describes how this architecture was implemented and Section 6.5 presents the testbed and the performed tests, and analyzes the obtained results as well. Finally, Section 6.6 concludes this chapter with providing interesting topics for future work in this area.

6.2

WiMAX Architecture

This section starts by describing the overall WiMAX architecture standardized by the WiMAX Forum, typically referred to as the WiMAX Network Reference Model (NRM). Then, we succinctly summarize the salient characteristics of IEEE 802.16 [4, 5, 11], the MAC and PHY layers used in WiMAX networks, emphasizing on the

6 Mobility Management Architecture for WiMAX Networks

255

data, control and management planes, as well as on mobility management procedures for mobile WiMAX. This section concludes with an introduction to the recently finalized IEEE 802.21 standard [6], which aims at optimizing mobility management procedures for handover scenarios, including networks based on IEEE 802.16, such as WiMAX.

6.2.1

The WiMAX Forum Network Reference Model

IEEE 802.16 defines a broadband wireless access technology for metropolitan area networks (MANs), supporting both fixed and mobile terminals, as defined in the IEEE 802.16-2004 [4] and IEEE 802.16e-2005 [5] standards, respectively. IEEE 802.16 specifies the air interface, including the MAC and the PHY layers. Nevertheless, it does deal with the radio access network functionalities. The standardization of the WiMAX access network is part of the WiMAX Forum. In particular, the WiMAX Forum Network Working Group (NWG), which is currently defining a high-performance All-IP (AIP) end-to-end network architecture to support fixed, nomadic, portable, and mobile users [3]. Besides defining an end-to-end IP framework, the WiMAX Forum also aims to ensure full interoperability between Base Stations (BSs) and Subscriber Stations (SSs)/Mobile Stations (MSs) from different WiMAX vendors. The WiMAX Forum introduced an architecture based on IEEE 802.16 by defining the NRM. The NRM, illustrated in Fig. 7.1, is a logical representation of the WiMAX network architecture, based on a set of functional entities and standardized interfaces, known as Reference Points (RPs). Table 6.1 summarizes the WIMAX NRM Reference Points. By implementing the WiMAX NRM, multiple

Table 6.1 WiMAX NRM reference points Reference Description Point R1 Interface between the MS and the BS: implements the IEEE 802.16e-2005 air interface. R2 Logical interface between the MS and the CSN: used for AAA, IP host configuration, and mobility management. R3 Interface between ASN and CSN: supports AAA, policy enforcement and mobility management; implements the tunnel between ASN and CSN. R4 Interface between ASN and ASN: used for MS mobility across ASNs. R5 Interface between CSN and CSN: used for internetworking between the home and visited networks. R6 Interface between BS and ASN-GW: implements intra-ASN tunnels; used for control plane signaling. R7 Interface between data and control plane in the ASN-GW: used for coordination between data and control plane in ASN-GW. R8 Interface between BS and BS: used for fast and seamless handovers.

256

S. Sargento et al. R2

WiMAX ASN

R1

Serving WiMAX BS

MS

WiMAX CSN

R6 R3

DNS

R8 CSN-GW R6

AAA

ASN-GW (Foreign Agent)

Target WiMAX BS

MIP (Home Agent)

DHCP

Fig. 6.1 The WiMAX forum network reference model

implementation options for a given functional entity are allowed, maintaining interoperability through the adoption and use of RPs R1-R8. As shown in Fig. 6.1, the WiMAX NRM defines three functional entities, namely, the Connectivity Service Network (CSN), the Access Service Network (ASN) and the MS. The MS is also referred to as the terminal equipment, and is responsible for establishing radio connectivity with the BS (via the R1 reference point). The ASN provides the connectivity between the MS and the IP backbone. The ASN is generally composed of several BSs connected to one or several ASNGateways (ASN-GW). The ASN-GW is the gateway for the ASN, aggregating all BSs’ information towards the CSN. The ASN is responsible for a set of important functionalities to provide radio connectivity to WiMAX subscribers, such as network discovery and selection, radio resource management, multicast and broadcast control, intra-ASN mobility, accounting, admission control, and Quality of Service. Furthermore, the ASN also performs relay functions to the CSN for IP connectivity establishment. Finally, the CSN complements the ASN by providing IP connectivity-related services. For example, the CSN operates Dynamic Host Configuration Protocol (DHCP), Domain Name Service (DNS) and Authentication, Authorization, and Accounting (AAA) servers, and oversees mobility management procedures.

6.2.2

The IEEE 802.16 Reference Model

Since IEEE 802.16 is a connection-oriented technology, all data transfers require the prior establishment of dedicated connections between the BS and the SS/MS. As detailed in [4], a connection, which is identified by a 16 bit Connection Identifier (CID), is a unidirectional mapping between the BS and the SS/MS MAC layer for

6 Mobility Management Architecture for WiMAX Networks

257

CS SAP CID A

Service Specific Convergence Sublayer

CID C

MAC SAP

Classifier CID D

MAC Common Part Sublayer MAC Layer

Control & Management NCMS Entity M-SAP

Security Sublayer

C-SAP

CID B

PHY SAP

PHY Layer

Physical Layer

Data Plane [IEEE 802.16d/e]

Control & Mgmt. Plane [IEEE 802.16g]

Fig. 6.2 IEEE 802.16 reference model (Data, control and management planes)

transporting the traffic of the corresponding service flow. During the SS/MS initialization process, three pairs of management connections (basic, primary and secondary) are established between the BS and the SS/MS, reflecting distinct QoS levels. The basic connection is used to transfer short, time-critical MAC management messages. The primary management connection transports longer, more delay-tolerant management messages. The secondary management connection is used to transfer delay-tolerant, standard-based management messages such as DHCP, Trivial File Transfer Protocol (TFTP) and Simple Network Management Protocol (SNMP). Moreover, a broadcast connection is configured by default to transmit MAC management messages to all receivers. Besides the aforementioned management connections, a multicast polling connection is also established so the SS/MS joins multicast polling groups and requests additional bandwidth. Finally, to satisfy the contracted services, transport connections are allocated to convey data packets in the air interface. As illustrated in Fig. 6.2, the MAC layer is divided in three sublayers, namely, the Service Specific Convergence Sublayer (CS), the MAC Common Part Sublayer (CPS) and the Security Sublayer. The CS is the first sublayer of the MAC layer, accepting higher layer MAC Service Data Units (SDUs) through the CS Service Access Point (SAP) and classifying them to the appropriate connection. The classifier is based on a set of matching criteria applied to each individual packet, consisting of specific protocol fields (e.g. IPv4/v6, Ethernet, Virtual Local Area Networks – VLAN, TCP/UDP ports), a classifier priority and a reference to a CID (see Fig. 6.2). Downlink classification is made at the BS whereas uplink classifiers are located in the SS/MS. When the classification process concludes, packets are delivered to the MAC Common Part Sublayer (MAC CPS) through the MAC Service Access Point (MAC SAP).

258

S. Sargento et al.

The MAC CPS is responsible for the core MAC functions, such as system access, bandwidth request and allocation mechanisms, connection establishment, contention resolution, and QoS management. QoS is catered for by associating each packet traversing the air interface with a service flow, as identified by the CID. According to IEEE 802.16-2004 [4], a service flow is a unidirectional flow of packets with a specific set of QoS parameters, such as latency, jitter, throughput and scheduling service. The scheduling service allows the IEEE 802.16 system to define transmission ordering and scheduling on the air interface, as well as how the SS/MS requests uplink bandwidth allocations. Five scheduling services are supported by IEEE 802.16: Unsolicited Grant Service (UGS), real time Polling Service (rtPS), extended real time Polling Service (ertPS) (only for IEEE 802.16e-2005 [5]), non real time Polling Service (nrtPS) and Best Effort (BE). Table 6.2 describes each one of the mentioned scheduling services. The Security Sublayer is the third and last sublayer from the MAC layer. This sublayer provides authentication and data encryption functions. The IEEE 802.16g-2007 [11] standard has been defined to efficiently integrate the IEEE 802.16 system with the higher layer control and management functionalities (see right-hand side of Fig. 6.2). In particular, 802.16g specifies the Network Control and Management System (NCMS) abstraction, which represents the higher

Table 6.2 IEEE 802.16 scheduling services Scheduling Target services service UGS Support real-time service flows that generate fixed size data packets on a periodic basis, such as VoIP.

rtPS

ertPS

nrtPS

BE

Support real-time service flows that generate variable size data packets on a periodic basis, such as MPEG video or streaming video. Support real-time service flows that generate variable size data packets on a periodic basis, such as VoIP with silence suppression. Instead of providing fixed allocations such as UGS, ertPS provides dynamic allocations. Support non-real-time service flows that require variable size data grants on a regular (but not strictly periodic) basis, such as high bandwidth FTP. Support service flows where no throughput or delay guarantees.

QoS parameters Maximum Sustained Traffic Rate (equal to the Minimum Reserved Traffic Rate) Maximum Latency Tolerated Jitter Request/Transmission Policy Maximum Sustained Traffic Rate Minimum Reserved Traffic Rate Maximum Latency Request/Transmission Policy Maximum Sustained Traffic Rate Minimum Reserved Traffic Rate Maximum Latency Request/Transmission Policy

Minimum Reserved Traffic Rate Maximum Sustained Traffic Rate Traffic Priority Request/Transmission Policy Maximum Sustained Traffic Rate Traffic Priority Request/Transmission Policy

6 Mobility Management Architecture for WiMAX Networks

259

layers entities (e.g. QoS and/or mobility management functions) that interoperate with the IEEE 802.16 system. Furthermore, the Control Service Access Point (C-SAP) and the Management Service Access Point (M-SAP) establish communication between an IEEE 802.16based system and NCMS entity(ies) for control and management purposes, respectively. The Management SAP (M-SAP) is used for less time-sensitive management plane primitives, such as system configuration, monitoring statistics, notifications, triggers and multi-mode interface management. Besides, the Control SAP (C-SAP) is used for more time sensitive control plane primitives, including handover, mobility management, security context management, radio resources management, subscriber and session management, and Media Independent Handover Function (MIHF) [6] services.

6.2.3

Mobility Management in WiMAX

Hard Handover (HHO) and Soft Handover (SHO) methods are supported by mobile WiMAX, defined in IEEE 802.16e. In the HHO method or “break-before-make” approach, the serving network link is broken before the handover execution is triggered and the target link is established. With respect to the SHO method or “make-before-break” approach, the target link is prepared and established before the serving network link is broken and the handover is executed. Although the SHO method provides continuous connectivity for the MS, it is much more complex and requires backbone communication between the serving and the target access technologies. In mobile WiMAX, two types of SHO are optionally supported, namely, the Fast Base Station Switching (FBSS) and the Macro Diversity Handover (MDHO). In the FBSS handover case, the MS is able to rapidly switch between several Base Stations without completing the entire network entry procedure. In the MDHO mode, the MS has simultaneous communication links with several Base Stations, enabling a fast handover to occur with minimum traffic degradation. Both mobile and network initiated handovers are supported in mobile WiMAX. For the former case (mobile initiated handover), the MS is responsible to trigger the handover initiation process, whereas for the network initiated handover scenario, the serving BS is in charge of triggering the mobility procedures. In order to provide a brief explanation about the handover procedures in Mobile WiMAX, we will present the example of a mobile initiated handover between BSs located in the same ASN (intra-ASN handover), as illustrated in Fig. 6.3. Before proceeding with the intra-ASN handover example, it is important to clarify that IEEE 802.16e-2005 [5] only defines the air interface procedures, including the MAC management messages that are exchanged between the Base Station and the MS. All the communication procedures in the backbone network are not covered by IEEE 802.16e-2005 itself. To overcome this gap, the WiMAX Forum has specified a network protocol [3] to establish the communication in the backbone network between the serving,

260

S. Sargento et al.

R1 (1, 5)

WiMAX ASN

R1 (2, 6) Serving WiMAX BS

R6

CSN-GW

MS

R8 (4)

R8 (3, 7) ASN-GW R6 Target WiMAX BS

Fig. 6.3 WiMAX intra-ASN mobile initiated handover

candidate and target Base Stations (reference point R8, illustrated in Fig. 6.3), as well as between the Base Stations and the ASN-GWs (reference point R6, also illustrated in Fig. 6.3). The intra-ASN handover scenario shown in Fig. 6.3 is composed of both IEEE 802.16e-2005 [5] MAC management messages (R1 reference point) and WiMAX Forum backbone messages (in this case, the R8 reference point). Initially, the MS is connected to the serving BS and receives the Mobility Neighbor Advertisement (MOB-NBR-ADV) MAC management message from the serving BS (Fig. 6.1), indicating that a WiMAX BS is available in the surrounding area. As a consequence, the MS triggers the handover preparation phase and sends the Mobility Mobile Station Handover Request (MOB-MSHO-REQ) MAC management message to the serving BS (Fig. 6.2) with a list of the candidate BSs for the handover. Thereafter, the serving BS sends a Handover Request (HO-REQ) backbone message to the candidate BSs (:3), Fig. 6.2, including the required QoS parameters to satisfy the MS currently running services. The candidate BSs reply with a Handover Response (HO-RESP) backbone message to the serving BS via the R8 reference point (:4), indicating whether the required QoS resources are available or not. The serving BS collects all responses and sends a Mobility Base Station Handover Response (MOB-BSHO-RESP) MAC management message to the MS (:5) with the list of recommended candidate BSs. Finally, the mobility decision algorithm of the MS selects the target BS (from the provided list of candidate BSs) and notifies the serving BS about the handover execution using the Mobility Handover Indication (MOB-HO-IND) MAC management message (:6). Finally,

6 Mobility Management Architecture for WiMAX Networks

261

the serving BS notifies the target BS that the MS will execute handover using the Handover Confirm (HO-CNF) backbone message (:7).

6.2.4

IEEE 802.21 Media Independent Handover Overview

The recently finalized IEEE 802.21 standard [6] aims to optimize the mobility management procedures in heterogeneous access environments. Towards this aim, it defines a Media Independent Handover (MIH) framework, which provides standardized interfaces between the access technologies and the mobility protocols from the higher layers in the protocol stack. The envisaged heterogeneous environment is illustrated in Fig. 6.4. A multi-operator, multi-technology network employing WiMAX, Wi-Fi and 3GPP UMTS/LTE is shown, including the IEEE 802.21 Point of Attachment (PoA) and Point of Service (PoS). PoA is the access technology attachment point, whereas PoS is the MIH entity that communicates with the multimode terminal. IEEE 802.21 introduces a new entity called MIH Function (MIHF) which hides the specificities of different link layer technologies from the higher layer mobility entities. Several higher layer entities, known as MIH Users (MIHUs) can take advantage of the MIH framework, including mobility management protocols, Non-PoS ¬ ® PoS MN ¬ ® PoS

WiMAX MIIS Server MIH Non-PoS

WiMAX MIH PoA

MIH PoS

IEEE 802 Operator A Home Core Network

Wi-Fi Wi-Fi MIH PoA

Non-PoS ¬ ® PoS MN ¬ ® PoS

MIIS Server MIH Non-PoS

MIH PoS

3GPP Operator B Visited Core Network

MN MIH

LTE LTE MIH PoA

UMTS UMTS MIH PoA

Fig. 6.4 MIH in a heterogeneous access network

262

S. Sargento et al.

such as Fast Mobile IPv6 (FMIPv6) [12], Proxy Mobile IPv6 (PMIPv6) [13] and Session Initiation Protocol (SIP) [14], as well as the other mobility decision algorithms. In order to detect, prepare and execute the handovers, the MIH platform provides three services: l

l

l

The Media Independent Event Service (MIES) provides event reporting such as dynamic changes in link conditions, link status and link quality. Multiple higher layer entities may be interested in these events at the same time, so these events may need to be sent to multiple destinations. The Media Independent Command Service (MICS) enables MIHUs to control the physical, data link and logical link layer. The higher layers may utilize MICS to determine the status of links and/or control a multimode terminal. Furthermore, MICS may also enable MIHUs to facilitate optional handover policies. Events and/or Commands can be sent to MIHUs within the same protocol stack (local) or to MIHUs located in a peer entity (remote). The Media Independent Information Service (MIIS) provides a framework by which a MIHF located at the MS or at the network side may discover and obtain network information within a geographical area to facilitate handovers. The objective is to acquire a global view of all the heterogeneous networks in the area in order to optimize seamless handovers when roaming across these networks. Figure 6.5 illustrates the IEEE 802.21 MIH Framework.

SIP L3

PMIP

FMIP

MIHU

MIH Events

L2.5

MIHU

MIH Commands

MIHF

MIES

MIH_SAP

Information Services

Remote MIH Events & Commands

MIHF

MIIS

MICS

MIH_NET_SAP Link Commands Link Events L2

Information Services

Link Layers

UMTS

WiMAX

Wi-Fi

Link Layers LTE

Wireless Access Technologies

Fig. 6.5 MIH framework

MIH_LINK_SAP

6 Mobility Management Architecture for WiMAX Networks

263

The following section introduces a QoS-aware mobility management architecture that capitalizes on IEEE 802.21 for inter-technology handovers for WiMAX networks.

6.3

QoS-Aware Mobility Management: A Use Case for the WEIRD System

The WEIRD overall architecture follows the WiMAX Forum NRM [3] and is illustrated in Fig. 6.6. The WiMAX resources dynamic control is handled by the coordinated action of the control plane modules located in the MS and in the ASN Gateway, while the CSN control plane manages the interaction between the WiMAX access network and the core network. Finally, the application servers and the AAA server are situated in the CSN as recommended by the WiMAX Forum. Quality of Service and mobility are managed in a coordinated way at the control plane level through the inter-communication and the combined processing of the

Fig. 6.6 WEIRD architecture

264

S. Sargento et al.

Connectivity Service Controller (CSC) modules, located in each segment of the NRM, and their interaction with the physical layer and the application layer. It is worthwhile to highlight that QoS and mobility management are strictly correlated and interdependent in WiMAX networks. QoS in WiMAX links is supported by a connection-oriented approach. Each WiMAX connection is associated with a set of downlink or uplink Service Flows characterized by a specific profile that defines both the scheduling class and the QoS parameters (bandwidth, priority, latency, jitter). In this architecture, each application session is associated with a set of endto-end flows with a QoS description based on the traffic type and application requirements. Two main issues must be considered in order to ensure the same QoS level along the full end-to-end path, including the WiMAX access segment, and independently of user mobility. Firstly, the end-to-end path can include several domains, like Diffserv or Intserv domains, supporting heterogeneous underlying network technologies and QoS guarantees, with different parameters and detail levels. The uniformity of the QoS level in this scenario, even with the bounds imposed by the different technologies, can be obtained using a multi-domain QoS signaling protocol, like QoS NSIS Signaling Layer Protocol (NSLP) [9] and mapping the QoS description in a set of specific parameters as defined by the QoS metric supported in each domain [15]. The QoS parameters carried in the QSPEC (QoS specification) are opaque to the QoS NSLP and are defined in the QoS model of the specific network technology, so that they can be interpreted only by the Resource Management Functions (RMF) of the NSIS peers located in each domain. In case of multi-domain path, the NSIS node located at the ingress of each domain is in charge of the translation between the received QSPEC and the internal QSPEC. This QSPEC is based on the QoS parameters supported by the underlying technology of the specific domain, and it can be easily understood by the RMF of each internal NSIS node. Figure 6.7 shows an example with a scenario including a WiMAX access network and a Diffserv domain in the core network. The QSPEC Wimax domain

Wimax QSPEC

BS SS

ASN-GW Internal NSIS node

Edge NSIS node Wimax domain Diffserv domain Diffserv QSPEC

Edge NSIS node Diffserv domain

QoS NSLP signaling Edge NSIS node Diffserv domain

Fig. 6.7 Multi-domain QoS NSLP signaling

6 Mobility Management Architecture for WiMAX Networks

265

carried in the QoS NSLP messages follows the Diffserv QoS model in the Diffserv domain and the WiMAX QoS model in the WiMAX domain. In particular, at the WiMAX access domain, the QoS description is based on a set of Service Flows characterized by the scheduling class and the WiMAX QoS parameters which are more suitable for the current session. These Service Flows are reserved and activated during the session setup phase using the network-initiated procedure defined as mandatory in the IEEE 802.16 d/e specifications [4, 5]. In this case, the Service Flow creation is initiated by the BS and controlled by the corresponding ASN-GW. With respect to mobility, the user must be able to move between different access networks. The mobility management architecture must be able to support not only handover between WiMAX BSs (located in the same ASN or in different ASNs), but also handover between different access technologies, like Wi-Fi and WiMAX networks. A further scenario, presented in detail in Section 6.3.2, is an intertechnology handover between Ethernet and Wi-Fi with the WiMAX network as the backhaul access technology. All these types of handovers must be transparent for the user and the same QoS level must be maintained during the entire length of the active sessions involved in the handover, if this handover is authorized through the user contracts. Therefore, QoS and mobility need to be managed together and independently on the specific link layer technology, through an interaction between the control plane and the lower layers of the involved access technologies. Notifications about the current link status can be processed by the Mobility Manager modules at the MS, ASN and CSN, in order to manage handovers through procedures that are completely transparent for the application layer. In particular, handover management in WiMAX networks involves the automatic reconfiguration of the wireless links through a “make-before-break” approach: the “make” phase includes the creation and the activation of new Service Flows in the segment between target SS and target BS, and the “break” phase includes the deletion of the preexistent Service Flows on the old wireless link. While this procedure can be controlled entirely by the ASN-GW in case of serving and target BS located in the same ASN, if the user moves between two different ASNs, the global handover management must be coordinated at the CSN level and the two ASN-GWs are just in charge of the resource control in the related WiMAX segments.

6.3.1

Integrated QoS and Mobility Management

As discussed earlier, the twofold interaction between the control plane and the application plane, on the one hand, and between the control plane and the transport plane, on the other, is of great importance. While the former allows the acquisition of the application QoS requirements for the resource control during the session setup and tear-down phases, the latter enables the control plane to modify the resource allocation during the handover.

266

S. Sargento et al.

During the session setup phases, the CSCs interact with the service layer in order to retrieve information from the applications, regarding the traffic type and the required QoS parameters. In particular, two different approaches can be adopted to support both legacy and IP Multimedia Subsystem (IMS), like applications based on the Session Initiation Protocol and Session Description Protocol (SIP/SDP) [14, 16] signaling. As shown in Fig. 6.8, in the first case [17], the QoS signaling follows the host-initiated approach. The CSC located at the MS (CSC_MS) communicates with a module, called WEIRD Agent (WA), in charge of obtaining parameters such as required bandwidth, maximum latency and jitter from the applications. CSC_MS coordinates end-to-end QoS signaling, using the NSIS framework, translating the application QoS requirements to a QSPEC adopted in the WiMAX NSIS model [15] and initiating the end-to-end signaling towards the ASN, the CSN and the core network. In the case of IMS-like applications (Fig. 6.9), the QoS signaling follows the network-initiated approach and it is strictly connected to the application layer SIP/SDP signaling [14, 16]. The SIP Proxy located at the CSN intercepts the SIP signaling between the SIP User Agents and extracts the session description from the SDP messages. The QoS parameters are forwarded to the CSC located at the ASN (CSC_ASN), through a set of Diameter [18] messages describing the media flows included in the sessions, where they are translated into WiMAX parameters. In this case the QoS NSIS signaling follows the edge-to-edge model since it is initiated and controlled by the CSC_ASN. For both legacy and IMS-like applications, WiMAX resource reservations are handled by the ASN-GW through the interaction of the CSC with the lower planes. In particular, at the link layer level the Resource Control (RC) module manages all

Fig. 6.8 Host-initiated QoS signaling and resource reservation

6 Mobility Management Architecture for WiMAX Networks

267

Fig. 6.9 QoS signaling and resource reservation for IMS-like applications

the WiMAX technology related functionalities, like Service Flow creation, modification and deletion, enforcing the QoS decisions on the WiMAX system through a set of technology dependent adapters [19]. Mobility management is based on the MIP protocol [20] and the IEEE 802.21 [6] framework that enables MIH defining an abstract layer for the unified interaction between the upper layer entities, called MIHU, and lower layer entities based on different technologies through a common interface. Figure 6.5 illustrates the abstractions introduced by IEEE 802.21. In the architecture (see Fig. 6.6), the mobility management framework includes several instances of the MIHF, located at each segment of the NRM. MIHF hides all technology dependent messages and procedures from the MIHUs exposing only a set of standardized interfaces that can be used to exchange common messages like events, commands or notifications among local or remote MIHUs and Link Layer Controllers (LLC). In the specific architecture, the following SAP are used to manage the handovers in the architecture: l l

MIH_LINK_SAP between the lower layers and the MIHF MIH_SAP between the MIHF and the MIHUs

MIHF also handles the exchange of the MIH protocol messages between remote MIHUs and LLCs providing three different services: MIES, MICS and MIIS. The MIH Events originate from the LLC and include information about the link layer, for example, the respective link status. The MIH Commands originate at the MIHUs and are used to convey the handover decisions. The transport of the MIH protocol messages between remote MIHF peers is supported by the NSIS framework through the Media Independent Handover NSLP (MIH NSLP), proposed by

268

S. Sargento et al.

the authors in [21], and Generic Transport Internet Signaling (GIST) protocol [8]. The MIH NSLP was developed as an extension to the NSIS framework in order to transport the MIH protocol messages. There are two main reasons to sustain this approach. First, the IEEE 802.21 proposed standard does not specify any protocol for message exchange, providing only the requirements for such protocol, namely, security and reliability. Second, QoS signaling, which is tightly coupled with mobility, is performed through the QoS-NSLP. In this context, the use of the NSIS framework to support both QoS and mobility process becomes the natural choice, since it fulfills the requirements for MIH message exchange between remote entities. As previously referred, the mobility management architecture, illustrated in Fig. 6.10, includes a Mobility Manager (MM) instance, acting as a MIHU and strictly connected with the related CSC, located on each NRM segment (MS, ASNGW, CSN), and a LLC located on the MS. The LLC is in charge of monitoring the link condition (signal level for Wi-Fi and WiMAX links, connected/disconnected cable for Ethernet). In case of link status variation, the related MIH Event is created and sent to the local MIHF through the MIH_LINK_SAP. Here, it is delivered to the registered MIHUs, both local and remote MMs, through the MIH_SAP. The MIH Events are used by the MM to update their internal status and detect new imminent handovers. In this case, the MM located at the MS searches for the availability of a new target link to be established and requests a new resource reservation to the associated CSC. The entire procedure is performed jointly by the CSC and the MM. While the MM manages the link status and is able to take decisions about the handover executions, the CSC handles the sessions at the control plane and controls the resources for the associated traffic flows. Following

Fig. 6.10 Mobility management architecture

6 Mobility Management Architecture for WiMAX Networks

269

the approaches used for resource control in the session setup and tear-down phases, the handover procedure and the wireless link reconfiguration are controlled by the MM located at the MS for host-initiated sessions and by the MM located at the ASN for IMS-like applications. When a mobile node moves between different ASNs, the entire procedure is controlled by the MM located at the CSN, which takes the final handover decision. However, the actual resource reservation is still performed by the CSC_ASN. Handover decisions are finally notified to the lower layers using the MIH Commands delivered to the LLC.

6.3.2

QoS-Aware Inter-technology Handovers for WiMAX

This subsection presents the sample scenario for inter-technology QoS-aware mobility, designed, implemented and deployed in the WEIRD project. It involves a vertical handover between Ethernet and Wi-Fi with WiMAX as the backhaul access technology, as shown in Fig. 6.11. A laptop with two different network interfaces (Ethernet and Wi-Fi) plays the role of the MS and it is initially connected via Ethernet cable to a LAN backhauled by an IEEE802.16d Subscriber Station (SS), the serving SS. When the user moves from its desk, it unplugs the cable and the laptop is connected through the Access Point to the Wi-Fi network, backhauled by another IEEE 802.16d SS, the target SS. The two SSs are located in the same ASN and can be connected to the same BS or to different BSs controlled by the same ASN-GW following the intra-ASN WiMAX mobility model. In our prototype, some applications are running on the laptop while it is connected via the Ethernet segment. The corresponding sessions are activated with the associated resources reserved along the end-to-end path. In particular, a set of WiMAX Service Flows has been activated between the serving SS and the Home Network

IEEE 802.16d SS - 1 Serving SS

MS Ethernet Connection

IEEE 802.16d BS ASN-GW

Application server

CSN AAA Server

Wi-Fi Access Point Foreign Network MS

Wi-Fi Connection

IEEE 802.16d SS - 2 Target SS

Fig. 6.11 Inter-technology mobility scenario

WiMAX ASN - CSN

270

S. Sargento et al.

associated BS during the session setup phase when the applications have been initialized. Each current Service Flow is characterized by the WiMAX parameters and the scheduling class more suitable for the QoS requirements of the active services. When the user moves to the Wi-Fi network, the same QoS level must be assured for each running application, so new Service Flows must be created and activated in the WiMAX segment between the target SS and the associated BS. This mobility scenario includes both intra-technology and inter-technology handover: the inter-technology handover involves the Ethernet and the Wi-Fi technology, but since they are backhauled by two different WiMAX SS when the MS moves, an intra-technology handover from the serving WiMAX SS to the target WiMAX SS must be performed. Figure 6.12 presents the handover signaling diagram applied to legacy applications, characterized by host-initiated sessions, for the scenario described in the previous paragraph. When the user starts a legacy application, the resource reservation procedure is triggered by the WEIRD Agent and the end-to-end QoS NSIS signaling is initiated as described in Section 6.3.1. As a result, a set of Service Flows are created at the ASN between the serving SS and the connected BS in order to assure the required

RC, GA, RSA

CSC_ASN

CSC_MS MM@MS

MIHF

MS Serving SS

MIH_LU (Eth)

SF

ETH

WA, LLC LU (Eth )

BS

Target SS Wi-Fi AP

LU (Wi-Fi)

Initialization and session setup

LLC

MIH_LU (Wi-Fi) QoS RES_REQ NSIS QoS RES _REQ RES_REQ RES_RESP

Create SFs at Eth side NSIS QoS RES _RESP

WA

QoS RES _RESP

LGD (Eth)

MS

MIH_LGD (Eth)

Serving SS SF

ETH

BS Target SS Wi-Fi AP

Handover preparation phase

SF

NSIS HO RES _REQ Create SFs at RES_RESP Wi-Fi side NSIS HO RES_RESP MIH_Link_Action

LLC

RES_REQ

Link_Action Serving SS

MS

Target SS Wi-Fi AP

SF

SF

Handover execution phase

Mobile IPv4

LD (Eth )

Serving SS

MIH_LD (Eth)

ETH

MS

Target SS Wi-Fi AP

Handover conclusion phase

NSIS HO DEL _REQ DEL_REQ DEL_RESP

SF

Delete SFs at Eth side NSIS HO DEL _RESP

Fig. 6.12 Signaling diagram for QoS-aware handover

6 Mobility Management Architecture for WiMAX Networks

271

QoS. It should be noted that the path including the Ethernet connection is automatically selected by the CSC located at the ASN, since the Wi-Fi connection is configured with a lower priority value and the network connected through SS#1 is considered the home network. At the beginning of the handover process, two Link_Up events are triggered by the LLC in order to notify the MIHF that both the Ethernet and the Wi-Fi connections are available. This event is forwarded to the registered MMs, in particular to the local MM (located at the MS) and to the remote MM (located at ASN and CSN), using MIH NSLP messages. Since for legacy applications the handover procedures are managed by the local MM, the MMs located on the other NRM entities and the related signaling are not shown in Fig. 6.12 and will not be further considered. When the Link_Up messages are received, the MM updates the internal state machine about the available network connections and is able to select the connection to be used for new sessions. The user can interact with the LLC in order to unplug the Ethernet cable and move to the Wi-Fi connection. The LLC detects that the Ethernet connection is going down and sends a Link_Going_Down event to the MIHF located at the MS that forwards it to the registered MMs. The MM at the MS internal state machine is updated again and, as consequence, an imminent handover is detected. The handover preparation phase is triggered in order to reserve the new resources before the Ethernet cable is unplugged. The MM at the MS selects the Wi-Fi link as the network connection to be used after the handover according to the current status of the internal machine and notifies this decision to the CSC at the MS. Here a new NSIS QoS signaling is triggered to update the resource reservation for the existing sessions and create new Service Flows (target SFs) in the segment between Home Agent (HA) located in the ASN-GW and the selected Foreign Agent (FA). The NSIS response message notifies the CSC at the MS that resources have been allocated between the target SS and BS and that they can be used by the data traffic flows after the handover. At this point the MS can move from the home network to the foreign network where it will be able to maintain the same level of QoS, so the MM at the MS can start the handover execution procedure through the MIH Link_Action command. When the user unplugs the Ethernet cable from the laptop, the Wi-Fi network interface starts the MIP registration with the FA, and the MIP tunnel between the FA and the HA at the ASN is established. Data traffic is carried through the Wi-Fi link and is mapped to the new Service Flows between the target SS and BS on the WiMAX link, ensuring the QoS level originally required by the active applications. The resources previously allocated between the serving SS and BS are released during the handover completion phase. When the Ethernet cable is unplugged, the LLC sends a Link_Down event, forwarded by the MIHF located at the MS to its MM. The CSC at the MS, as responsible of the dynamic control of sessions and resources, is in charge of handling the deletion of the old Service Flows for the existing sessions and initiates the related NSIS QoS signaling towards the CSC at the ASN.

272

6.4

S. Sargento et al.

Implementation

In this section, we take a closer look at the design and implementation of the components developed in order to demonstrate the feasibility of the mobility management architecture introduced in the previous section. More specifically, we present the MIHF module implemented, as well as the LLC, NSIS framework (regarding QoS signalling and MIH transport mechanisms), and at last, we present the details of the implementation of CSC, with its own MM.

6.4.1

MIHF Framework

The MIHF, located at the MS and ASN, is the central unit of the IEEE 802.21 architecture. It provides communication with lower layers through MIH_LINK_ SAP, with upper layers through MIH_SAP, and with remote MIHFs through MIH_NET_SAP, using the MIH protocol. Its implementation follows a set of steps, which are necessary for the proper integration with the architecture. Figure 6.13 shows the main features of the MIHF entity implemented. As can be seen, initially it is required to configure all the topology of the network with the important information of different units. Then, the creation of communication means (sockets) with all modules is performed. Each MIHF provides a set of

LLC

MIH_LINK_SAP

Configure the topology; Create sockets;

Listen Sockets

MIH Users

MIH_SAP

Receive message

Identify the message

Remote MIHFs

MIH_NET_SAP

Forward the message (locally or remotely) MIHF

Fig. 6.13 MIHF implementation overview diagram

6 Mobility Management Architecture for WiMAX Networks

273

events (Link_Up, Link_Down and Link_Going_Down), commands (Link_Action), information services (MIH_Get_Information) and elements of control (MIH_Register and MIH_Subscribe). After the MIHU has subscribed to the services provided by the MIHF, it will become associated with these services. Then, the MIHF is able to receive any message from MIHUs, LLCs or remote MIHFs, and will have the adequate behaviour depending on the message received. The main functions that must be performed by MIHF are the following: l

l

l

Interaction with Link Layer Controller: MIHF must have the ability to receive events from the LLCs, process these messages, identify the associated event and forward it to the MIHUs (local or remote) that subscribe it. Interaction with Media Independent Handover User: MIHF must have the ability to receive messages coming from MIHUs who want to register or subscribe its events. The registration or the subscription may be requested by a local MIHU or by a remote MIHU. Moreover, MIHF must generate and submit their response to these requests. Overall interaction: Additionally to the direct interactions specified above, MIHF must also have the ability to receive the MIH_Link_Action.request and send the respective answer to all MIHUs or LLCs who signed to this type of command.

6.4.2

Link Layer Controller

The aim of the LLC is to implement a link information collector independent of the specific hardware, vendor, or GNU/Linux kernel. For this, Linux natively provides convenient ways for application layer software to gather link specific information from the kernel and directly from network device drivers without modifications to both of them. LLC, illustrated in Fig. 6.14, constantly monitors the network link states and, based on this information, it provides events through an Event Trigger module to the registered MIHF. For simplicity, in the examined scenario, LLC provides only Link_Up, Link_Down and synthetically generated Link_Going_Down

MIHF TCP Event Trigger

Generic Link State Monitor

Link-specific Information Monitor

Kernel

Fig. 6.14 Schematic of LLC

802.3 Initiate

802.11

274

S. Sargento et al.

events. The monitored link types are Ethernet (IEEE 802.3) and Wi-Fi (IEEE 802.11). Link states are identified in the Generic Link State Monitor (GLSM) by observing the operation status of access network interfaces. After each link is operationally up and its link type has been identified, GLSM initiates the Link-specific Information Monitor (LSIM) which acquires link-specific information. For instance, LSIM can obtain Access Point information for Wi-Fi accesses. This information is gathered using ioctl system calls. Since the Link_Up message content, specified in IEEE 802.21, optionally includes the MAC address of an access router (gateway) assigned to the link, LSIM also verifies and monitors the connectivity with the access router. In this study, LLC is configured to trigger a Link_Up event only if a mobile node has L2 connection with both the AP and the assigned access router.

6.4.3

NSIS: QoS and MIH Transport

The NSIS framework includes different software modules, namely MIH NSLP for the transport of MIH messages, QoS NSLP to accomplish QoS signalling, and GIST, which acts as the transport layer. Figure 6.15 shows the different modules of the NSIS framework and their interconnections. All NSIS modules employ TCP/IP sockets for their communication. Also the interface with upper layers, such as MIHF and applications is based on TCP/IP sockets. The MIH NSLP module implements different functionalities to achieve the MIH message transport between remote MIHF entities. The processing of MIH messages, including message parsing, enqueueing and mapping, is one of the first operations performed by the MIH NSLP. The mapping process corresponds to a map between a destination ID determined by the MIHF, to an IP address, which is required by GIST. The instructions for the delivery process, such as the type of transport required (assured delivery) and the destination address, are included in the Message Routing Information (MRI) object. Such instructions are fundamental for MIH User MIHF

Application

MIH NSLP

QoS NSLP

GIST

Fig. 6.15 NSIS framework decomposition

NSIS

6 Mobility Management Architecture for WiMAX Networks

275

the operation of GIST. Therefore, at this second stage, the serialization of MRI is performed. These operations are executed in the nodes initiating the MIH messages, namely the Mobile Station. Afterwards, the destination nodes or the nodes intercepting such messages perform the reverse operations. For instance, the MIH NSLP installed on the ASN node performs deserialization of the messages received from GIST, in order to obtain the MIH payload. QoS NSLP, as a NSLP for QoS signalling, performs operations similar to the ones described above for the MIH NSLP, such as serialization and deserialization of MRI. Nevertheless, besides some functional similarity, the QoS NSLP module implements other more complex operations, such as soft-state management. State maintenance is performed within specific messages and it is used to maintain sessions that are active and that require resources. In the architecture presented, QoS NSLP northbound interface is the CSC. GIST, as the transport layer of the NSIS framework, is the foundation for the proper message signalling, both in terms of QoS and MIH functionalities. GIST can work using unreliable and reliable modes of operation. For instance, if the MRI object specifies the need for reliable transport, GIST will establish message associations between peer entities. Being a generic transport protocol, GIST has the capacity to support a NSLP for different purposes, such as the MIH NSLP and QoS NSLP described above.

6.4.4

MM and CSC

As described in Section 6.3, each segment of the WiMAX network (MS, ASN and CSN) is managed by a CSC, with its own MM. In the architecture, the role of the CSCs is to coordinate the WiMAX network control plane. In particular, the main functions of the Connectivity Service Controller are the following: l l

l

l

Manage sessions with the associated QoS signalling mechanisms. Coordinate authorization procedures in cooperation with the Diameter (AAA) server [18]. Manage WiMAX resources, which are dynamically created, activated, updated, and deleted during the session setup and tear-down. Control the resource update during the handover phases following the “makebefore-break” model. This reconfiguration is coordinated together with the MMs that handle all mobility-related procedures, monitoring the status of the network connections of the controlled MSs and making handover decisions.

The high-level flow chart of the dynamic session management and the corresponding control of the WiMAX resources is illustrated in Fig. 6.16. WiMAX resources are firstly created and activated during the session setup. The first step of the handover towards a foreign network is triggered by the LLC. At this stage, the LLC issues a Link_Going_Down message and then new WiMAX Service Flows are created in the target wireless segment. The end of the

276

S. Sargento et al.

Fig. 6.16 Session and resource dynamic control

New Session ® Activate Resources Link Going Down ® Create target SFs ACTIVE SESSION

HO MAKE

Link Down ® Remove serving SFs Session Termination ® Remove existing resources

Session Termination ® Remove existing resources END

handover execution is signalled by the Link_Down message. Then, the old WiMAX resources between serving SS and BS are released. Finally, during the tear-down phase, all the existing Service Flows associated to the session are removed.

6.4.5

QoS Signalling and Resource Reservation

This section describes the procedures for QoS signalling and resource reservation for legacy and SIP-based applications. CSC at the MS is the main coordinator for legacy applications that follow the host-initiated model for QoS signalling. In this case, the CSC at the MS acts as the NSIS signalling initiator and creates the first QSPEC, translating the application requirements (as received from the WEIRD Agent) in a set of parameters based on the WiMAX QoS model [15]. In this scenario, the ASN at the GW is a generic QoS NSIS node located inside a WiMAX domain on the end-to-end path, and the CSC at the ASN acts as a local RMF: it receives requests from the NSIS layer and, if authorized, allocates the resources in the WiMAX link through the RC. In case of SIP-based applications based on the SIP/SDP protocol, networkinitiated QoS signalling is adopted. The CSC at the ASN is the main coordinator for both the WiMAX resource reservation and the edge-to-edge NSIS signalling through the CSN and the core network. The media flow description included in the SDP messages is retrieved by the Application Function (AF) located on the SIP Proxy of the CSN and transferred to the CSC at the ASN through Diameter messages. Here, resources are allocated and activated on the WiMAX link following the two-phase activation model defined by the IEEE802.16d/e specification.

6 Mobility Management Architecture for WiMAX Networks

277

During the first phase of the SIP call, a set of provisioned SFs is created with the maximum level of QoS required by the codecs specified in the SDP messages. At the end of the SIP negotiation phase, the SFs are activated with the QoS parameters corresponding to the audio/video codecs agreed by the two SIP clients. The mechanism for resource allocation on the WiMAX link follows the same network-initiated model for both legacy and SIP-based applications, and it is entirely managed by the CSC_ASN at the ASN-GW. The procedure consists of the following steps: l

l

l

l

Mapping between the QoS parameters included in NSIS or SDP messages and a set of WiMAX Service Flows QoS authorization through an interaction between the CSC at the ASN and the Diameter AAA server located in the CSN Admission Control (AC) for the ASN segment resources, including the WiMAX SFs Interaction between the CSC at the ASN and the RC in order to create and activate the SFs

6.4.6

Mobility

The management of the handovers and the corresponding resource updating follow the same approach of the resource reservation control during the session setup and teardown phases. In case of legacy applications, characterized by host-initiated sessions, all handover procedures are handled by the joint action of CSC and MM both at the MS, while all the processing is controlled by the CSC and MM at the ASN for SIP-based applications. In particular, the MM handles the status of the network connections of the corresponding MS through the events received from the local MIHF module and detects imminent handovers, taking active decisions and triggering the resource updating. This reconfiguration is controlled by the CSC, which is able to retrieve from its internal state all required information about the running sessions involved in the planned handover and, in particular, the WiMAX resources associated to the active data traffic flows that must be re-allocated. In this section, we present in detail the handover procedure introduced in Sections 6.3.1 and 6.3.2. This scenario is characterized by an inter-technology handover between Ethernet and Wi-Fi, with WiMAX as the backhaul access technology. Legacy applications with host-initiated sessions are considered, therefore all handover procedures are handled by the MM at the MS. The status of the MS network connections is handled by a mobility manager submodule, called Node Manager (NM), and is updated whenever a new MIHF message is received from the local MIHF module. The network interface information managed by the NM is shown in Table 6.3. The Network Interface ID is the unique identifier associated to the network interface, while the Network Interface Type defines the network technology. In this scenario only Ethernet and Wi-Fi technologies are considered. However, other types of network interfaces can be

278

S. Sargento et al. Table 6.3 Network interface information handled by the node manager Network interfaces information Network Interface ID Network Interface Type (ETH/Wi-Fi) MAC Address IP Address Priority (0/1) Status (Up/Going-Down/Down) Active (True/False) With traffic (True/False)

START Ethernet: UP T Wi-Fi: UP

Ethernet MIH Link Going Down ® NSIS HO Make Request

NSIS HO Make Response ® Ethernet MIH Link Action Command

Wi-Fi MIH Link UP ® null Ethernet: UP T Wi-Fi: D

Ethernet: GD Wi-Fi: UP T Ethernet MIH Link Down ® NSIS HO Break Request

NSIS HO Break Response ® null

Ethernet: D Wi-Fi: UP T

Ethernet: UP T Wi-Fi: D

NSIS HO Break Response ® null

Wi-Fi MIH Link Down ® NSIS HO Break Request Ethernet: UP T Wi-Fi: GD NSIS HO Make Response ® Wi-Fi MIH Link Action Command Ethernet: UP WiFi: GD T

Ethernet: GD T Wi-Fi: UP

Ethernet: D Wi-Fi: UP T Wi-Fi MIH Link Going Down ® NSIS HO Make Request

Ethernet MIH Link Up ® null Ethernet: UP Wi-Fi: UP T

Fig. 6.17 MM at the MS finite state machine transitions

specified if supported by the MS device. The Status field includes an enumerated value (Up/Going-Down/Down) that describes the current status of the network interface as notified by the received MIHF messages. In order to decide which connection is active when more network interfaces are “UP”, the integer value of the Priority field is considered. This value is commonly configured with “0” for Ethernet (higher priority) and with “1” for Wi-Fi (lower priority). As a consequence, when both the interfaces are “UP”, the Boolean Active field is true for the Ethernet connection. Finally, the Boolean “With traffic” field is adopted to mark the network connection that is currently used to carry data traffic flows. Figure 6.17 shows the finite state machine for a MM handling two network connections: Ethernet with higher priority and Wi-Fi with lower priority. Each state describes the current status of the network interfaces: up (UP), going-down (GD) or down (D). The network connection which is currently used to carry data traffic is marked with the “T” flag. The right side of the figure shows the handover procedure from the Ethernet to the Wi-Fi connection, while the left side shows the handover from the Wi-Fi to the Ethernet connection.

6 Mobility Management Architecture for WiMAX Networks

279

Resource updating for mobility follows the “make-before-break” approach and is completely transparent to the application layer. When the MM receives a new MIHF Link_Going_Down message for a connection that is currently carrying data traffic, it initiates the procedures for resource reconfiguration. As a result, the CSC at the MS sends an NSIS request to the CSC at the ASN with the Handover Make command. In its turn, the CSC at the ASN creates a set of new Service Flows on the target segment. At the end of the resource allocation process, the MM sends a MIHF Link_Action message to the MIHF so that the handover procedure can proceed. The MIHF Link_Down message announces the end of the handover procedure and the availability of the new L3 connection at the end of the Mobile IP (MIP) registration. Resources between serving SS and BS are released through the NSIS request with the Handover Break command, and data traffic is mapped on the Service Flows over the WiMAX link between target SS and BS. The handover management has been designed in order to minimize the amount of WiMAX resource reconfiguration procedures: if in a given instant the Wi-Fi network connection with lower priority is carrying data flows and the Ethernet interface with higher priority comes back to the “UP” status, the traffic of the existing sessions is maintained on the first interface. The MM only triggers the resource reconfiguration if a Wi-Fi interface Link_Going_Down message is received.

6.5

Testbed Evaluation

This section describes the empirical evaluation of the proposed mobility management architecture prototype. The experimental scenario was depicted in Fig. 6.11. The testbed includes modules that implement the CSN, ASN and MS functionalities. Under the ASN, we install a real, commercial-of-the-shelf (COTS) WiMAX BS directly connected to the ASN-GW. Two WiMAX SSs are connected to the BS creating a Point-to-Multipoint topology. The MS can be connected to SS1 by Ethernet and to SS2 by Wi-Fi. The video server is located in the CSN and the video client in the MS. To carry out the handover between Ethernet and Wi-Fi, we use MIPv4. Since the beginning, we already know that MIPv4 has inherent latency problems because of the packet tunnelling between the Home Agent and the Foreign Agent. However, we use it as demonstrator because of its simplicity. Due to MIPv4 we have to define the Home Network and the Home Agent, the Foreign Network and the Foreign Agent, and the Mobile Node. Table 6.4 shows the testbed components. The goal of this testbed is to evaluate the effectiveness of WiMAX technology in a mobility scenario (handover between Ethernet and Wi-Fi, backhauled by WiMAX), regarding QoS reservation in WiMAX links and user’s authentication mechanisms. Initially the MS is connected to Ethernet. Therefore, it is necessary to reserve in the BS-SS1 WiMAX link, two Service Flows (one uplink and one downlink) with an allocated bandwidth of 512 Kb/s, to handle a video stream.

280

S. Sargento et al. Table 6.4 Testbed components Base Station/Subscribers Stations (RedMAX)

Desktops MS’s Laptop

Frequency band Channel bandwidth PHY WiMAX Downlink ratio Uplink & downlink modulation for SSs BS-SSs distances Ubuntu 8.04 (Linux) Ubuntu 8.04 (Linux)

3.5 GHz 7 MHz 16d, OFDM 56/44 64QAM 15 m

Table 6.5 Architecture modules involved in the scenario for the testbed Module Description VLC Client and Server Application CSC_MS Connectivity Server Controller located on the MS Link Layer Controller (LLC) Monitoring the state of MS Network Interfaces WEIRD Agent (WA) Resource Reservation provisioning for Legacy Applications MIHF MIH Function (subset of the IEEE 802.21 specification) NSIS framework NSIS Signalling Generic Adapter Service Flow management Redline equipment Specific Adapter Resource Controller (RC) Control the Resource Reservation on the ASN Gateway AAA Server Authentication, Authorization and Accounting Administration Console Control Plane Monitoring CSC_ASN Connectivity Service Controller located on the ASN Gateway CSC_CSN Connectivity Service Controller located on the CSN MIP (Home Agent, Foreign Agent enabling the Mobile IP protocol Agent, Mobile Node)

While the user is watching a video received through the concatenated WiMAX/ Ethernet link, he/she decides to unplug the Ethernet cable and connect to the Wi-Fi AP. This automatically triggers a vertical handover procedure between Ethernet and Wi-Fi, which will initiate the Service Flows reservation in the BS-SS2 WiMAX link. After this process, the user performs the handover and can continue to watch the video by the composed WiMAX/Wi-Fi link, experiencing the same videoquality. A brief summary of the architecture modules involved in the scenario and their main functionalities is provided in Table 6.5. The performance of the proposed mobility architecture is addressed in the remainder of this section. After a brief overview of the overall handover process, the procedures to provide QoS-aware mobility are analyzed evaluating the internal processing times of the involved modules and the communication times between remote entities for both QoS and MIHF signalling. We repeated each measurement run ten times and report for the median value of the obtained times.

6 Mobility Management Architecture for WiMAX Networks

6.5.1

281

Handover Process Overview

As discussed in Section 6.3, the handover process follows the “make-before-break” model and consists of three main sequential phases: handover preparation, handover execution and handover conclusion. The first step includes all the procedures to configure the target segment, while during the handover execution the data path towards the foreign network is established. Finally, existing resources on the old path are removed during the handover conclusion phase. In order to ensure the same level of QoS for the active applications when the user moves to the foreign network, the configuration of the target resources during the handover preparation requires the coordination between the entities handling mobility (Mobility Manager) and session control (CSC), as well as the exchange of NSIS messages between MS and ASN for the creation of new Service Flows. Similar procedures have been adopted during the handover conclusion to delete the old resources. The resource reconfiguration phase (handover preparation) is based on IEEE 802.21 technology. It enables to provide QoS-aware mobility and have no impact on the handover execution phase, which is based on MIPv4 [20], and includes the creation of the MIP tunnel between the Foreign Agent and the Home Agent, located in the ASN-GW. Table 6.6 shows the duration of the three handover phases: 736 ms for the handover preparation, 4,199 ms for the handover execution and 655 ms for the handover conclusion. It is important to highlight that the most critical values from the user’s point of view are the handover preparation and the handover execution times since they can lead to some interruptions to the data flows and a consequent degradation of the QoS at the application level. On the other hand, the handover conclusion is completely transparent and has no consequence on the QoS perceived by the user. The handover execution is measured as the time interval between the instant when the Ethernet interface stops receiving the video streaming and the moment when the Wi-Fi interface starts receiving it. The higher value of the handover execution time is due to the MIPv4 and the problems associated with the routing of the packets to the Care-of-Address of the MS. It is known that MIPv6 has improvements relatively to MIPv4, including among others, the redundancy of Foreign Agent entities in the network as the mobile host itself can handle the Foreign Agent functionalities, a native solution for the triangle routing, dynamic address auto-configuration also for Care-of-Addresses and improved security. Nevertheless, for simplicity reasons, and since our focus was in the IEEE 802.21 and QoS integration, MIPv4 was used in this demonstrator. Table 6.6 Handover processing time Handover phase Time (ms) Preparation 736 Execution 4,199 Conclusion 655

282

6.5.2

S. Sargento et al.

Mobility Manager Initialization

The initialization phase allows the system to be configured with all the required information about the WiMAX network topology and the available network connections for the controlled mobile nodes. While the location of the BSs and their connections with the ASN can be statically configured, other data like the configuration of fixed IEEE 802.16d SS and MS must be dynamically updated. Moreover, when a new device joins the network, there is the need to notify the MMs of the WiMAX network about the status of the network connections between the fixed SS and the connected mobile nodes. During the start-up procedure, the LLC of the MS sends an Ethernet Link_Up event and a Wi-Fi Link_Up event to the MIHF at the MS, and here the messages are forwarded to the registered MMs. Since MIHF at the MS receives the events until it forwards them to the MM at the MS, it takes nearly 215 ms to process the events. The processing time is approximately 4 ms at the MM at the MS and 5 ms at the MM at the ASN, including the updating of the internal status for both network interfaces of the mobile node.

6.5.3

Resource Reconfiguration During HO Preparation and Conclusion

During the handover preparation phase, the resources on the WiMAX/Wi-Fi link are allocated to support the handover of the MS from the home network to the foreign network. A set of service flows is created and activated on the WiMAX link between the target SS and BS, in order to support the media flows of the active video streaming. After the handover execution, during the conclusion phase, the resources reserved in the WiMAX/Ethernet link are released so that they can be available for future connections (consequently, the service flows allocated between the serving SS and BS are also deleted). Figure 6.18 shows the full processing time for handover preparation and conclusion phases in case of mobile-initiated handovers. In both cases the main component is the duration of the NSIS bidirectional communication between the MS and the ASN-GW (87% for handover preparation and 94% for handover conclusion). The MM and CSC at the MS and the CSC at the ASN processing time are detailed in Fig. 6.19 for the handover preparation and the handover conclusion phases. The MS processing time (nearly 70 ms for the handover preparation and 25 ms for the handover conclusion) is due to the CSC at the MS module, where the MM acts as the coordinator of the entire handover procedure. During the first step of the handover preparation, the MM updates the internal state machine with the current status of the MS Ethernet connection and triggers the handover, while the CSC

6 Mobility Management Architecture for WiMAX Networks

283

ASN Processing Time + Cross Layer MS Processing Time MSASN Communication

800 700

Time (ms)

600 500 400 300 200 100 0 HO Preparation

HO Conclusion

Fig. 6.18 Processing and communication time for mobile-initiated handover

100 90 80 Time (ms)

70

CSC/MM@MS - Session update and Link Action

60 50

CSC_ASN - Target resources creation

40 30 20 10

CSC_MS - Session update CSC_MS - Target resources computation

CSC_ASN - Old resources deletion CSC_MS - Old resources retrieving

MM@MS - Ho make triggering

0 HO Preparation

MM@MS - HO break triggering

HO Conclusion

Fig. 6.19 Processing time for HO preparation and conclusion

retrieves the QoS requirements of the stored sessions and computes the new resources to be allocated in the target link for each of them. The corresponding NSIS QSPEC is sent to the NSIS module to initiate the signalling to the CSC at the ASN through the WiMAX link. The processing time for this first step is approximately 37 ms (13 ms for the MM and 24 for the CSC). In the last step of the handover preparation, when the new resource reservation has been notified by the NSIS response, the CSC at the MS updates the status of the stored sessions, while the MM sends the Link_Action message to the MIHF module, with a total processing time of nearly 33 ms. The conclusion phase is very similar for the first step

284

S. Sargento et al.

when the deletion of the old Service Flow is triggered, while the last step is characterized by a lower processing time since it does not include the exchange of a MIHF command. The processing time at the ASN-GW includes both the processing of the CSC at the ASN and the cross-layer modules. It is approximately 25 ms for the handover preparation and 15 ms for the handover conclusion). These modules do not take any active handover decision, since they are only in charge of the WiMAX resources reconfiguration for the active sessions, as specified in the received QSPEC. In particular, during the handover preparation phase the new Service Flows are allocated and activated in the WiMAX segment towards the target SS, while during the handover conclusion the Service Flows are deleted over the serving WiMAX link. NSIS communication time between the MN and the ASN is the highest for both handover preparation and conclusion phases. This is due to the NSIS processing time, namely the message association performed by GIST between the first nodes on the preparation phase and changes due to the mobility which affect the NSIS framework as stated in [22]. Besides, all the signalling between the MS and ASN QoS NSIS Entities (QNE) nodes crosses WiMAX links, on which the message delay is approximately 30 ms. With respect to the MIHF at the MS, during the handover preparation phase the processing time to forward the Link_Going_Down event received from the LLC is nearly 215 ms (as we can see in Fig. 6.20). The MIHF processing time at the MS, after receiving the Link_Action from the MM and before sending the message to the LLC, is 145 ms. During the handover conclusion phase, the MIHF at the MS takes approximately 215 ms to process the Link_Down event received from the LLC. It is noticeable that the internal processing time of the MIHF is much smaller than in the CSC modules both in the MS and ASN. The MIHF, when properly configured and initialized, just has to forward events and commands to the MIHUs and LLCs.

LD_Eth

Case of Study

LGD_Eth LU_WiFi LU_Eth LA LD_Eth

MIHF_ASN

LGD_Eth LU_WiFi LU_Eth

MIHF_MS (remote communication)

LD_Eth LGD_Eth

MIHF_MS (local communication)

LU_WiFi LU_Eth

0

50

100

150

200

250 Time (us)

Fig. 6.20 MIHF processing time

300

350

400

450

500

6 Mobility Management Architecture for WiMAX Networks

285

Finally, the communication time between the MIHF and the CSC is around 750 ms on each direction.

6.5.4

MIH Transport Mechanism

As stated before, the MIH transport mechanism relies on the NSIS communication facilities, namely GIST, to guarantee the transport of messages, and MIH NSLP, to instruct GIST on the delivery process. The processing time of GIST includes the parsing of MIH messages received from the MIH NSLP, the forwarding to the next peer, and the refresh mechanisms to keep the associations. Since the MS acts as the initiator, GIST has a higher processing time when compared with the ASN. This observation is due to the decision process on the transport protocol (UDP or TCP), as well as on the message association mechanism required by GIST. On the ASN side, the GIST processing time is nearly 7.5 ms. Table 6.7 depicts the processing time of MIH NSLP which includes different processes in the MS and the ASN. At the MS, the MIH NSLP processes the messages received by the MIHF (MIH messages to be transported to a remote MIHF), and due to the messages received, the MIH NSLP instructs GIST on the delivery process through the MRI serialization. The MIH Processing at the MS takes approximately 25 ms and includes the parsing of MIH messages, in order to map the destination ID to an IP address (required by the forwarding process of GIST). The MRI Serialization at the MS side lasts nearly 15 ms. At the ASN side, the MIH NSLP processes the messages received from GIST and performs the necessary processing to deliver the MIH messages to the remote MIHF. This process takes around 7.5 ms. All these values are small and do not compromise the handover efficiency. In a remote communication, when MIHF at the MS forwards the events sent by LLC, the MIHF processing time is nearly 310 ms. These results stem from the fact that MIHF receives the event until it forwards them to the NSIS at the MS in order to make a remote communication between the MIHF both at the MS and ASN. Then, when the MIHF at the ASN receives the MIH messages from NSIS (also at the ASN), it has a processing time of nearly 300/400 ms, in order to forward them to CSC (as we can see in Fig. 6.20). The results presented in this section have shown the processing times of the architecture modules involved in the different phases of the vertical handover between Ethernet and Wi-Fi links, detailing the cost associated with the mechanisms to achieve a “make-before-break” approach. Table 6.7 MIH NSLP processing time Phase MIH Processing MRI_Serialization Receive Processing

Time (ms) 25.88 15.23 8.08

286

6.6

S. Sargento et al.

Conclusions

As mobile communication becomes widespread over a wide set of wireless technologies, there is the need for mechanisms that support seamless inter-technology handovers. Moreover, given the user requirements for next generation applications, inter-technology handover mechanisms must be developed while maintaining adequate levels of quality of service. This chapter has presented a solution that addresses this twofold problem based on a QoS-enabled architecture for mobility management in WiMAX networks. Mobility management in heterogeneous environments, with inter-technology handovers, can be substantially improved by the use of a unifying framework such as the Media Independent Handover, described in IEEE802.21. With such an approach, the details of the underlying technologies become transparent to the upper layers, allowing a smoother support of vertical handovers. Seamless handovers call for a “make-before-break” solution, where resources are reserved in the target network before the connection to the serving network is broken. In the mobility management architecture described in this chapter, the IEEE 802.21 Media Independent Handover standard was integrated with the Next Steps in Signalling framework for achieving quality of service signalling in the inter-technology mobility scenario. The chapter introduced a use case developed in the context of the WEIRD project. Implementation aspects about the integration of the Media Independent Handover standard with the Next Steps in Signalling framework within the mobility management architecture were also detailed. The results obtained from an empirical evaluation of a prototype implementation of the proposed architecture of the main mobility and quality of service mechanisms indicate that the solutions presented in this chapter are able to seamlessly integrated these two worlds of QoS and mobility management in a media independent way. Acknowledgment Part of this work was conducted within the framework of the IST Sixth Framework Programme Integrated Project WEIRD (IST-034622), which was partially funded by the Commission of the European Union. Study sponsors had no role in study design, data collection and analysis, interpretation, or writing the book chapter. The views expressed do not necessarily represent the views of the authors’ employers, the WEIRD project, or the Commission of the European Union. We thank our colleagues from all partners in WEIRD project for fruitful discussions.

References 1. ITU-T, General Principles and General Reference Model for Next Generation Networks. Recommendation Y2011 (Oct 2004) 2. ITU-T, General Overview of NGN. Recommendation Y.2001 (Dec 2004) 3. WiMAX Forum, WiMAX End-to-End Network Systems Architecture Stage 2-3: Architecture Tenets, Reference Model and Reference Points, Release 1, Version 1.2 (Jan 2008)

6 Mobility Management Architecture for WiMAX Networks

287

4. IEEE 802.16 WG, IEEE Standard for Local and Metropolitan Area Networks. Part 16: Air Interface for Fixed Broadband Wireless Access Systems. IEEE Standard 802.16-2004 (Oct 2004) 5. IEEE 802.16 WG, Amendment to IEEE Standard for Local and Metropolitan Area Networks. Part 16: Air Interface for Fixed Broadband Wireless Access Systems – Physical and Medium Access Control Layer for Combined Fixed and Mobile Operation in Licensed Bands. IEEE Standard 802.16e-2005 (Dec 2005) 6. IEEE 802.21 WG, IEEE Draft Standard for Local and Metropolitan Area Networks: Media Independent Handover Services. IEEE P802.21/D10.0 (Apr 2008) 7. R. Hancock, G. Karagiannis, J. Loughney, S. Van den Bosch, Next Steps in Signaling (NSIS): Framework, IETF RFC 4080 (June 2005) 8. H. Schulzrinne, R. Hancock, GIST: General Internet Signalling Transport, IETF NSIS WG Internet-Draft (July 2008) 9. J. Manner, G. Karagiannis, NSLP for Quality-of-Service Signaling, IETF NSIS WG InternetDraft (Feb 2008) 10. G. Martufi, M. Katz, P. Neves, M. Curado, M. Castrucci, P. Simoes, E. Piri, K. Pentikousis, Extending WiMAX to new scenarios: Key results on system architecture and testbeds of the WEIRD project, in Proceedings of the Second European Symposium on Mobile Media (EUMOB) (Oulu, Finland, July 2008) 11. IEEE 802.16 WG, IEEE Standard for Local and Metropolitan Area Networks: Part 16: Air Interface for Fixed Broadband Wireless Access Systems; Amendment 3: Management Plane Procedures and Services. IEEE Standard. 802.16g-2007 (Dec 2007) 12. R. Koodli, Fast Handovers for Mobile IPv6, IETF RFC 4068 (July 2005) 13. S. Gundavelli, K. Leung, V. Devarapalli, K. Chowdhury, B. Patil, Proxy Mobile IPv6, IETF RFC 5213 (Aug 2008) 14. J. Rosenberg, H. Schulzrinne, G. Camarillo, A. Johnston, J. Peterson, R. Sparks, M. Handley, E. Schooler et al., SIP: Session Initiation Protocol, IETF RFC 3261 (June 2002) 15. N. Ciulli, G. Landi, M. Curado, G. Leao, T. Bohnert, F. Mitrano, C. Nardini, G. Tamea, A QoS model based on NSIS signalling applied to IEEE 802.16 networks, in 2nd IEEE International Broadband Wireless Access Workshop (Las Vegas, Nevada, USA, Jan 2008) 16. M. Handley, V. Jacobson, C. Perkins, et al., SDP: Session Description Protocol, IETF RFC 4566 (July 2006) 17. M. Castrucci, I. Marchetti, C. Nardini, N. Ciulli, G. Landi, An architecture for the QoS management in a WiMAX network – Analysis and design, in International Conference on Late Advances in Networks (Paris, France, Dec 2007) 18. P. Calhoun, J. Loughney, E. Guttman, G. Zorn, J. Arkko, Diameter Base Protocol, IETF RFC 3588 (Sept 2003) 19. P. Neves, T. Nissil€a, T. Pereira, I. Harjula, J. Monteiro, K. Pentikousis, S. Sargento, F. Fontes, A vendor-independent resource control framework for WiMAX, in Proc. 13th IEEE Symposium on Computers and Communications (ISCC) (Marrakech, Morocco, July 2008), pp. 899–906 20. C. Perkins, IP Mobility Support, IETF RFC 2002 (Oct 1996) 21. L. Cordeiro, M. Curado, P. Neves, S. Sargento, G. Landi, X. Fu, Media Independent Handover Network Signaling Layer Protocol (MIH NSLP), IETF NSIS WG Internet-Draft (Feb 2008) 22. T. Sanda, X. Fu, S. Feong, J. Manner, H. Tschofening, Applicability Statement of NSIS protocols in Mobile Environments, IETF NSIS WG Internet-Draft (July 2008)

Chapter 7

Radio Resource Management in WiMAX Networks Ljupcˇo Jorguseski and Ramjee Prasad

Abstract This chapter presents the general Radio Resource Management (RRM) problem in wireless access networks and gives performance evaluations for different downlink resource (sub-carrier) allocation algorithms in WiMAX TDD systems based on OFDMA wireless access. This is particularly important in downlink due to the traffic asymmetry and due to the information availability at the base station regarding the quality requirements for a particular user. The comparative performance analysis of the different resource allocation algorithms is using the average cell/user throughput and user throughput versus distance from the reference base station as performance metrics. For the comparative performance evaluation, we have resorted to MATLAB simulations. Additionally, a semi-analytical approach is proposed for the downlink throughput performance estimation of the algorithms that do not rely on the channel quality information feedback from the mobile users. The analysis shows that Proportional Fair and Soft partitioning/re-use algorithms are good candidates for downlink resource allocation in WiMAX TDD systems and that the proposed semi-analytical approach accurately estimates the throughput performance of the reuse schemes.

7.1

Introduction

This chapter addresses the radio resource management (RRM) in OFDMA based cellular networks such as WiMAX. One of the major RRM task is the allocation of the time-frequency resource units for the different users in a way that user specific quality thresholds are satisfied and at the same time the overall throughput of the system is maximized. This is particularly important in downlink due to the traffic L. Jorguseski (*) TNO Information and Communication Technology, Department – Access Network Technologies, Postbus 5050, 2600 GB, Delft, The Netherlands e-mail: [email protected]

R. Prasad and F.J. Velez, WiMAX Networks, DOI 10.1007/978-90-481-8752-2_7, # Springer ScienceþBusiness Media B.V. 2010

289

290

L. Jorguseski and R. Prasad

asymmetry and due to the information availability at the base station regarding the quality requirements for particular user. This chapter also presents simulation and semi-analytical approaches for estimating WiMAX throughput performance. The chapter is organized as follows. Section 7.2 presents a general definition of the RRM problem for a cellular system. The definition of resource units (RUs) is presented in Section 7.3. The different downlink RU allocation algorithms investigated in this chapter are presented in Section 7.4. The simulation and semianalytical approach for evaluating WiMAX system performance are presented in Sections 7.5 and 7.6, respectively. The chapter is finalized with the conclusions and recommendations in Section 7.7.

7.2

General Radio Resource Management (RRM) Problem Definition

Let us assume a WiMAX cellular network with M active mobiles in the service area denoted with M = {1, 2, . . . M}. Denote with B = {1, 2, . . . B} the set of all base stations (BSs) used to provide the necessary cellular coverage. The radio propagation link between each pair BS and mobile subscriber can be characterized with the link gain gi,j, where i ¼ 1, 2, . . ., B represents the index of the BS and j ¼ 1, 2, . . . M denotes the index of the mobile user. The link gains gi,j include all propagation phenomena such as path-loss, shadowing, and multi-path fading. This is schematically presented in Fig. 7.1. The radio link gain g links the received power level Prx at the input of the receiver with the transmitted power Ptx at the output of the transmitter as follows: Prx ¼ Gt Gr gPtx 1

2 g21 g1M

B gB1

g22

g12

g2M

gB2 gBM

g11

1

(7.1)

2

M

Fig. 7.1 Schematic view of the radio link gains in a wireless cellular network

7 Radio Resource Management in WiMAX Networks

291

Here, Gt and Gr are the transmitter and receiver antenna gains, respectively. In order to describe the instantaneous radio conditions in the cellular system under investigation we can use the link gain matrix G that is defined as: 0

g11 g12 ::: g1M

1

B g g ::: g C B 21 22 2M C G¼B C @ :::: A

(7.2)

gB1 gB2 ::: gBM The gain matrix G is dynamic; the dimension M is changing depending on the number of active mobile users and each element gij changes with the mobile’s movement. Note that in general we have different gain matrix G for uplink and downlink (e.g. if different frequency bands are used for uplink and downlink). The elements of this matrix are describing the individual radio link gains between the pairs BS and mobile subscriber. Assume that each BS from the set B has a limited set of waveforms C that can be allocated to the mobile users and C ¼ {1, 2, . . .C} is the numbered set of these waveforms. Note that with a waveform we refer to the combination of: l l

l

Modulation scheme such as BPSK, QPSK, 16QAM, etc. [1]. Channel coding such as convolutional coding [2], turbo coding [3], etc. with a certain coding rate. Multiple access specific signal form such as the combination of time-slot(s) and frequency in TD/FDMA system (e.g. GSM, GPRS or EDGE), channelisation/ spreading code in DS-CDMA systems (e.g. UMTS), or a set of sub-carriers in the OFDM signal in WiMAX and E-UTRAN wireless cellular systems.

The radio resource management problem can be defined as follows. Taking into account the link gain matrix G as presented in Section 7.2 the RRM algorithms in the given wireless cellular system handle the following assignments: l l l

Assign one or more access point from the set B. Assign uplink and downlink waveform from the set C. Assign transmitting power for the BS and for the mobile.

These assignments should be such that the following quality requirements in uplink and downlink are satisfied: ul SINRul i;j  gj

(7.3)

dl SINRdl i;j  gj

(7.4)

dl Here, gul j and gj are the quality targets for the individual users in uplink and downlink, respectively. Note here that the uplink and downlink quality requirements for the SINR can be derived from the link-level evaluations of the particular

292

L. Jorguseski and R. Prasad

radio bearer and depend on the desired communication service (e.g. voice, video, web-browsing, etc.), radio propagation characteristics, and the transmission/ reception algorithms implemented at the BS and the user terminal. Radio resource management (RRM) aims at optimal “management” or (realtime) “allocation” of the available radio resources among the currently active users given the radio network infrastructure and its capacity. The optimal management in the context of RRM is governed by the QoS levels that the wireless operator provides to its customers and operator’s policies. The operator’s policies are driven by the operator’s business plan. This requires a complex and dynamic interaction between the different RRM algorithms (e.g. admission/congestion control, radio link adaptation, scheduling, handover, etc.) in the network to achieve the desired QoS levels according to the operator’s policy, given the current system conditions. These interactions between the RRM algorithms should jointly facilitate the overall optimization goal. The RRM algorithm usually aims at the optimization of an objective function given the quality related conditions (7.3) and (7.4). The objective functions can relate to the desired policy by the operator. Here are few examples for the operator policies: l

l

l

l

Serve as many users as possible: the RRM algorithms would aim at maximizing the number of concurrently active users given the target conditions (7.3) and (7.4). Maximizes the total achieved throughput in downlink and uplink: the RRM algorithm assignments would optimize the total sector throughput in downlink and uplink given the conditions (7.3) and (7.4). Minimize the downlink and uplink interference: the RRM algorithm assignments would minimize the used power in downlink and uplink (especially interesting for interference limited systems). The likelihood that the user throughput is lower than a minimum level Rmin should be lower than 5%. The RRM algorithm assignments would then aim at optimizing the assignments in such a way that the uplink/downlink throughput per user is higher than the pre-defined threshold Rmin with sufficient probability.

The RRM algorithms take also the responsibility for an efficient utilization of the available radio spectrum. The efficient usage is of crucial importance because radio spectrum is a scarce and costly resource. Measures must be taken to ensure this effectiveness, and this is handled at two stages – network planning and during network operation via RRM functions. Usually the extended set of (advanced) RRM functions is implemented at the BS. This is because the BS can perform and gather measurements that are used as input for the RRM functions and they additionally require significant processing. At new session set-up or handover request the admission control decides whether to admit the user to the BS. The admission/rejection decision is based on interaction with the load/congestion control, in order to assess the current load conditions, and with the scheduling and link adaptation in order to assess the achievable throughputs. The load control function monitors the interference and the current load in the BS and takes appropriate

7 Radio Resource Management in WiMAX Networks

293

actions (e.g. orders the admission control to not admit any new users) to resolve overload/congestion situations. The power and rate assignments are executed by the e.g. packet scheduling and radio link adaptation functions in order to satisfy the different QoS requirements for the new and ongoing sessions. These functions also interact with the admission/load control algorithms in order to adjust the scheduling order or the link adaptation. In the following, these important RRM algorithms are briefly explained.

7.3

Radio Resource Definition in WiMAX

In WiMAX system, OFDM transmission is scalable. The scalability can be made by changing the size of FFT and consequently the number of sub-carriers. The supported FFT sizes are 128, 512, 1024, and 2048, but only 512 and 1024 are mandatory for Mobile WiMAX. The 802.16e PHY supports Time Division Duplex (TDD), Full and Half Frequency Division Duplex (FDD) operations. However, the initial release of mobile WiMAX only includes TDD. Figure 7.2 shows the OFDMA frame structure for a TDD implementation. The data is configured into frame of 48 OFDM symbols. Each frame is divided into downlink (DL) and uplink (UL) sub-frames separated by Transmit/Receive and Receive/Transmit Transition Gaps (TTG and RTG). Each DL sub-frame starts with a preamble followed by the Frame Control Header (FCH), the DL-MAP, and the UL-MAP. In the DL, sub-channels may be assigned for different users. Frame duration is 5 ms. Each frame consists of 37 data symbols and 11 overhead symbols.

Fig. 7.2 WiMAX OFDMA frame structure

294

L. Jorguseski and R. Prasad

The 37 data symbols can be partitioned between the uplink and downlink according to a certain ratio, for example DL versus UL ratio of 28:9. For the RRM algorithm evaluations in this chapter the WiMAX TDD resource model consists of 35 OFDM symbols (28 data symbols and 7 overhead symbols)/DL sub-frame [5], and 840 useful sub-carriers/OFDMA symbol [4]. These 840 sub-carriers are organized into 30 RUs (28 sub-carriers/RU) [4]. There are two types of sub-carrier permutations for sub-channelization: diversity and contiguous [4]. The diversity permutation draws sub-carriers pseudo randomly to form a sub-channel such as FUSC (Fully Used Sub-carrier), PUSC (Partially Used Sub-carrier), and additional optional permutations. It minimizes the probability of using the same sub-carrier in adjacent cells or sectors. The contiguous permutation groups a block of contiguous sub-carriers to form a sub-channel such as AMC. The channel estimation in this case is easier than in the case of diversity permutation.

7.4

Downlink RRM Algorithms for WiMAX Systems

In a WiMAX cellular OFDMA system we have two-dimensional Frequency-Time (F-T) resource as presented in Fig. 7.3. In the frequency domain we have N subcarriers labeled f1 to fN. In the time domain we have OFDM symbols with duration Ts, grouped in Transmission Time Intervals (TTIs). The system allocates a set of sub-carriers per TTI (or multiple of TTIs) to a particular user that is referred to as Resource Unit (RU). Note that one or more RUs can be allocated to a particular user. This is indicated with the colored shaded areas in Fig. 7.3 with the assumption that the RU is allocated per TTI, each TTI contains eight OFDM symbols, and there are K resource units with different number of

Fig. 7.3 Downlink time-frequency resource in OFDMA cellular system; Ts is the OFDM symbol duration; TTI is the transmission time interval; gray shaded area is one OFDM symbol; different color per RU designates different user

7 Radio Resource Management in WiMAX Networks

295

sub-carriers per RU. The AMC scheme allocates one modulation and coding scheme for the whole RU and this is done per TTI (or multiple TTIs) basis. Denote with ‘m’ an arbitrary user in a WiMAX cellular system presented is communicating with ‘b’-th base station (b ¼ 1, 2, . . ., NB), sector ‘s’ (s ¼ 1, 2, 3) and has been allocated RU ‘r’ (r ¼ 1, 2, . . ., K). Then the signal-to-interference and noise ratio (SINR) for the allocated RU can be calculated as: SINRm;b;s ðrÞ ¼

PRU b;s ðrÞgm;b;s ðrÞ NB P 3 P i¼1 j¼1

PRU b;s ðrÞ gm;i;j ðrÞ

RU Nth ¼ kTWRU

PRU i;j ðrÞgm;i;j ðrÞ

þ

; j 6¼ s if i ¼ b RU Nth

is transmission power of the r-th RU from the ‘b’-th base station, and sector ‘s’ is the radio link gain (path-loss, shadowing, multi-path fading, and antenna gain) between user ‘m’ and sector ‘j’ from base station ‘i’ and for the resource unit ’r’. The dependency on the RU index is due to the frequency selective multi-path fading. is the thermal noise level for the RU, calculated from the Boltzman constant, receiver temperature, and RU bandwidth

The achievable bit-rate Rm,b,s(r) given the allocated RU ‘r‘ with bandwidth ‘Br’ depends on the allocated bandwidth and the achievable SINRm,b,s(r) that is Rm,b, s(r) ¼ f(Br, SINRm,b,s(r)). The RU allocation algorithm aims at finding the optimum solution according to a certain optimization criteria such as for example minimize the total downlink power, maximize the total downlink throughput in the system, etc. This optimization should be done given the conditions that all users achieve their minimum throughput requirement, the total power per base station does not exceed the physical maximum and that each RU is allocated to not more than a single user: ( min P

Total

¼

NB X 3 X

PTotal i;j

¼

i¼1 j¼1

NB X 3 X K X

) PRU r;i;j

i¼1 j¼1 r¼1

or ( max RTotal ¼

NB X 3 X

Ri;j ¼

i¼1 j¼1

Rm ¼

Mi;j X NB X 3 K X X i¼1 j¼1

K X

!) gm;r Rm;i;j ðrÞ

s.t:

m¼1 r¼1

gm;r Rm;i;j ðrÞrRmin m

(7.5)

r¼1

PTotal i;j bPmax

(7.6)

296

L. Jorguseski and R. Prasad Mi;j X

gm;r b1; 8r ¼ 1; :::; K

(7.7)

m¼1

Mi;j gm;r RU ¼ kTWRU Nth

is the number of users active in base station ‘i’ and sector ‘j’. is the integer variable equal to 1 if r-th resource unit is allocated to the m-th user; otherwise it is zero. is the thermal noise level for the RU, calculated from the Boltzman constant, receiver temperature, and RU bandwidth

Condition (7.5) ensures that the minimum QoS level (e.g. user bit rate) for the m-th user is satisfied. Condition (7.6) ensures that the physical limit for the total transmitted power per sector is not exceeded. Condition (7.7) ensures that only one user per RU has been allocated. It the optimization problem presented above we have a non-linear dependency between the allocated power and the achievable RU throughput that is Rm,b,s(r) ¼ f(Br, SINRm,b,s(r)). This non-linear dependency (usually logarithmic) results in a non-linear optimization problem. Additional complexity is the integer character of the allocation variables gm,r making the optimization problem mixed with real and integer variables. In order to avoid complex optimization solving of the RU allocation problem in the following sections several heuristic/practical downlink RU allocation algorithms are proposed.

7.4.1

Reuse-One

All available RUs can be allocated in a random fashion in each sector with reuse one. More than one users can be randomly selected for each RU. This is the simplest form of allocating the sub-carriers to different users and is considered as a reference case. Note that a user is allocated one or more RUs and that some users might not get any RU allocated.

7.4.2

Reuse-Three

The number of available resource units NRU is divided in three disjoint sets, which are consequently used in different sectors as presented in Fig. 7.4. Within one sector the users are allocated to RUs of this sector randomly as it is the case with the reuse-one. The division of sub-carriers into NRU total number of RUs is done in a continuous way in the frequency domain as presented in Fig. 7.5.

7 Radio Resource Management in WiMAX Networks

297

Fig. 7.4 Sub-carrier reuse three 2 2

3

1 3

2

1

2 1

2

3

2

Node-B

3

1 3

1

2 1

1 3

3

Target cell

1

NRU – 5

4

NRU – 2

Sector 1

f 2

NRU – 4

5

NRU – 1

Sector 2

3

6

NRU – 3

f NRU

Sector 3

f

Fig. 7.5 Sub-carrier divisions per RU for different sectors in frequency domain – reuse three

7.4.3

Soft Re-Use

All available sub-carriers are used in each sector as in case of Reuse-one, but the power allocated to sub-carriers is not equal as presented in Fig. 7.6. The number of sub-carriers with Pedge (dark color) is one third of the total number of sub-carriers, hence, the rest (2/3) will have power Pclose (light color). The users are classified into “close” (close to the cell center) and “edge” (close to the cell edge) by comparing the user’s average geometry Gavg with a pre-defined threshold Gth : Gavg ¼

NRU 1 X GðrÞ NRU r¼1

GðrÞ is the average geometry per RU and is given by: . PRU r;b;s Lb;s ; where j 6¼ s if i ¼ b GðrÞ ¼ N N . cell P sec tor P RU RU Pr;i;j Li;j þ Nth i¼1 j¼1

(7.8)

(7.9)

298

L. Jorguseski and R. Prasad

is the allocated transmission power to the rth RU from the bth base station PRU r;b;s and sector s. Li,j is the propagation loss including path loss, shadowing, antenna gain between user and the ith base station and sector j. RU is the thermal noise level for RU. Nth All users with Gavg b Gth are “edge” users, and users with Gavg > Gth are “close” users. All RUs are allocated with Pclose to close users and Pedge to edge users in the random fashion. A soft re-use named “re-use partitioning” presented in Fig. 7.7 is supported in WiMAX system. The user close to a base station can operate with all available sub-channels. In case of an edge user, each sector can operate with a fraction of all available

1

NRU – 5

4

NRU – 2

Sector 1

f 2

NRU – 4

5

NRU – 1

Sector 2

3

6

NRU – 3

f NRU

Sector 3

f

Fig. 7.6 Soft reuse – ‘edge’ sub-carriers (dark color) have higher power than ‘close’ sub-carriers (light color) NRU – 5

1 4

5

NRU

6

Sector 1

f

NRU – 4

2 4

5

NRU

6

Sector 2

f

NRU – 3

3 4

5

6

NRU

Sector 3

f

Fig. 7.7 Sub-carrier division in re-use partitioning in WiMAX system

7 Radio Resource Management in WiMAX Networks

299

sub-channels [6], for example, a fraction of 1/3 as in our model. The WiMAX case shows that the full load frequency reuse one is kept for “close” users to maximize spectrum efficiency, whereas fractional frequency reuse is used for “edge” users to improve their throughput as well as mitigate inter-cell interference among users.

7.4.4

Maximum C/I

In this case it is assumed that CQI information is available at the downlink subcarrier allocation algorithm at the NodeB as reported per RU from each mobile. This radio channel quality information is used in the allocation decision as illustrated in Fig. 7.8. Assume that the three different curves in Fig. 7.8 represent the normalized radio channel realization (i.e. the multi-path frequency selective power profile) for three different users.1 The maximum C/I sub-carrier allocation algorithm allocates the RUs Normalized − Pedestrian A 2.5 User 1 User 2 User 3

Power per subcarrier

2

1.5

1

0.5

0 0

50

100

150

200

250

300

350

400

450

500

Subcarrier number

Fig. 7.8 The maximum C/I based sub-carrier allocation algorithm for three users (512 subcarriers)

It is normalized in a sense that all users have average sub-carrier power E[P(k)] ¼ 1 over all subcarriers.

1

300

L. Jorguseski and R. Prasad

to the user that has the best radio channel conditions for that particular RU. This is illustrated in Fig. 7.8 with the different thickness of the curves. Note that all the subcarriers (RUs) between one and 48 and 123 and 251 are allocated to User 1 (thick solid line). Similarly, all sub-carriers where the channel conditions are the best for User 2 (User 3) marked with dot-dashed (dashed) line and allocated to this user. Note that User 3 will be allocated only a small part of the subcarriers between 49 and 122. This is due to the bad radio-conditions for User 3 with respect to the other two users. This means that users in bad radio conditions in this algorithm will “starve” for resources.

7.4.5

Proportional Fair (PF)

The PF scheduler calculates the ratio of the feasible rate to the average throughput for each user. The user with maximum ratio will be selected for transmission at the next coming time slot. In time slot t, the feasible rate of user k is Rk(t) and its average throughput is denoted by . Then user k* with maximum is served in time slot t. And the average throughput of each user is updated by [7]: 8  1  1 > > > 1  RðtÞ þ Rk ðtÞ <  tc k tc   Rðt þ 1Þ ¼  > 1 k > > 1 RðtÞ : tc k

k ¼ k (7.10) k 6¼ k

where tc is the time constant for the moving average.

7.5

Simulation Evaluation of Downlink Sub-Carrier Allocation Algorithms

For the performance evaluation of the different sub-carrier allocation algorithms we have used the system scenario with 19 three-sectored base stations as presented in Fig. 7.4. We have generated from ten to 30 users per sector in the three sectors of the central base station in order to avoid edge effects. We have assumed that all RUs are occupied in the surrounding base stations. A crucial modeling step is how to decide from the SINR calculated on the system level per RU the corresponding throughput (or BLER). Note that the BLER versus SINR performance is usually determined via link-level simulations with AWGN channels. Consequently, the appropriate modulation and coding scheme can be assigned that gives the maximum throughput for a given BLER target (e.g. BLER ¼ 10%). In this chapter we use the Effective Exponential SIR Mapping (EESM) approach [8], which is presented in Fig. 7.9. The EESM is especially applicable to the case of low/medium Doppler, that is when the channel can be assumed to be constant during a Transmission Time

7 Radio Resource Management in WiMAX Networks

301

Link level

System level

SIReff (G ; h) BLEPAWGN

BLEP (G ; h)

Mapping funct. SIReff ( . )

G,h

- Determine short-termaveraged Geometry - Generate instantanoeus channel

Link adaptation, Scheduling, ARQ, etc.

Throughput

Fig. 7.9 Effective Exponential SIR Mapping (EESM) methodology in OFDMA [8]

Interval (TTI). In the case of high Doppler (when the channel cannot be assumed to be constant over the TTI) the EESM cannot be applied as straightforwardly as for the case of low/medium Doppler. The reason is that, in this case, the effective SIR cannot be assumed to be constant during the TTI. Thus, a more complex mapping function may be needed for accurate performance estimation. The EESM works as follows: l

l

l

l

On system level, a UE exists within the simulated deployment. Based on the UE position, a short-time-average geometry G is calculated as the ratio between the desired signal level originated from the reference sector and the interference level originated from the surrounding cells. The geometry G includes the effect of distance-dependent path-loss, shadowing, and thermal noise. For each UE and each TTI, an instantaneous channel-impulse response hðtÞ is generated according to the average channel-delay profile e.g. Pedestrian B, Pedestrian A, etc. Based on the geometry G and the instantaneous channel-impulse response hðtÞ an effective SIR is calculated and used in the link-level evaluations. From the link-level simulations with AWGN channel the corresponding BLER is determined for a particular modulation and coding scheme, which in turn is used for the system level evaluations (see Fig. 7.9).

The link level results [9] (Tables 3.1–3.4) are used for the mapping between the effective SINR and the block error rate (BLER) for each AMC combination. Note here that these link level results were found as most appropriate publicly available results at the time of writing of the thesis. The link level performance depends heavily on the physical layer transmission and reception algorithms (e.g. MIMO techniques, interference cancellation techniques, etc.), radio channel knowledge (e.g. ideal or non-ideal channel estimation or feedback), etc. Therefore, the absolute performance results for the downlink throughput in the WiMAX system presented in this chapter have to be comprehended by having these reference link level results in mind. The interested reader should be aware that the absolute downlink throughput performance of WiMAX presented in this chapter will change if other reference link level

302

L. Jorguseski and R. Prasad

results are considered. However, the relative performance trends between the different allocation algorithms should still hold even if other link level results are used. The performance evaluation of the different downlink resource allocation schemes was done by MATLAB simulations. The wireless cellular system was modeled with one central three-sectored site surrounded with two tiers of interfering three-sector sites. The simulations were done for different number of users per sector randomly placed in the central site with uniform spatial distribution. The users are assumed continuously active in downlink (i.e. the full buffer approach). The geometry thresholds of 8 dB and 14 dB for the Soft Reuse and Partitioning Reuse allocation algorithms were used, respectively. These values were selected by trial simulations as a best compromise between satisfactory average sector throughput and reasonable throughput for cell edge users, as shown in Table 7.1, Figs. 7.10 and 7.11. Table 7.1 The sensitivity of geometry threshold Allocation algorithms Average sector throughput (Mbps) 2.9269 Gth1 ¼ 0 dB 2.5448 Gth2 ¼ 0 dB 6.1217 Gth1 ¼ 8 dB(selected) 5.9526 Gth2 ¼ 14 dB(selected) 8.6212 Gth1 ¼ 15 dB 7.1100 Gth2 ¼ 20 dB

Cell edge throughput (at 0.475 km) (Mbps) 0.0411 0.0398 0.0534 0.0632 0.0451 0.0509

Note: Gth1 and Gth2 are geometry threshold for Soft Reuse and Partitioning Reuse, respectively

Normalized User Throughput [bit/s/Hz]

Sensitivity of Geometry Threshold for Soft Reuse; Pedestrian B; WiMAX Gthr1 = 0 dB Gthr1 = 15 dB Gthr1 = 8 dB (selected)

0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

Distance [km]

Fig. 7.10 WiMAX user throughput versus distance for soft reuse, 30 users/sector

0.5

7 Radio Resource Management in WiMAX Networks

303

Normalized User Throughput [bit/s/Hz]

Sensitivity of Geometry Threshold for Partitioning Reuse; Pedestrian B; WiMAX Gthr2 = 0 dB Gthr2 = 20 dB Gthr2 = 14 dB (selected)

0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0

0

0.05

0.1

0.15

0.2 0.25 0.3 Distance [km]

0.35

0.4

0.45

0.5

Fig. 7.11 WiMAX user throughput versus distance for partitioning reuse, 30 users/sector

Although the selected geometry thresholds do not give us the maximum average sector throughput they perform better than the other geometry thresholds in terms of throughput for cell edge users. From Fig. 7.10 it can seen that at distances further than 0.275 km the selected geometry threshold outperforms the other in terms of user throughput versus distance. The other important system parameters are presented in Table 7.2. The normalized average sector and user throughput for WiMAX is presented in Figs. 7.12 and 7.13, respectively. The normalization is done with the system bandwidth. Note that the vertical bars presented in these figures are the 95% confidence intervals. It can be observed that the maximum C/I resource allocation algorithm has the best performance regarding the average sector and user throughput. This is because the RUs are allocated always to the user that has the best channel conditions at the moment of allocation. The Proportional Fair, Soft Re-use and Reuse-Partitioning have similar performance with regard to the average sector/user throughput. However, in a practical implementation it is expected that the proportional fair algorithm will perform better as it uses the channel quality feedback from the UEs and tries to achieve fairness by weighting the achieved throughput in the previous allocations. The soft reuse schemes in this study were already ‘optimized’ for this uniform spatial distribution of the end users by selecting an appropriate geometry threshold for deciding if the UE is ‘close’ or at the ‘cell edge’. In practice this could be difficult to achieve on the fly for different spatial distributions of the UEs. The reuse three has the worst performance as it uses only one third of the available bandwidth.

304

L. Jorguseski and R. Prasad

Table 7.2 The main system parameters for the simulations

Parameter fcarrier (GHz) fsampling (MHz) Bandwidth (MHz) Frame length (ms) Nr. OFDM sym. Cell radius (km) Total sub-carriers Used sub-carriers fspacing (KHz) Guard time (s) Symbol dur. (s) Number of RUs Total DL power (W) DL modulation Coding rates Antenna gain Antenna FTB ratio Link-level interf. Nth (dBm/Hz) ssh (dB) Shadow corr.

WiMAX 2.5 11.2 10 5 48 per frame 0.5 1,024 840 10.9375 11.43e-06 102.86e-06 30 40 QPSK, 16/64 QAM 1/2, 2/3, 3/4, 5/6 16 dBi 25 dB EESM 174 8.9 0.5

Average Sector Throughput vs Number of Users; Pedestrian B; WiMAX

Normalized Sector Throughput [bit/s/Hz]

1.6

1.4 Re-use 1 Re-use 3 Soft re-use Maximum C/I Proportional Fair Soft partitioning

1.2

1

0.8

0.6

0.4

0.2

5

10

15 20 Number of Users per Sector

Fig. 7.12 WiMAX sector throughputs with 95% confidence intervals

25

30

7 Radio Resource Management in WiMAX Networks

305

Average User Throughput vs Number of Users; Pedestrian B; WiMAX 0.16 Re-use 1 Re-use 3 Soft re-use Maximum C/I Proportional Fair Soft partitioning

Normalized User Throughput [bit/s/Hz]

0.14 0.12 0.1 0.08 0.06 0.04 0.02 0

5

10

15

20

25

30

Number of Users per Sector

Fig. 7.13 WiMAX User throughputs with 95% confidence intervals

The reuse one has better performance than the reuse three and its performance can be improved by either differentiation between close and edge users (e.g. soft reuse or reuse partitioning) or implementing channel aware resource allocation (e.g. proportional fair or maximum C/I). Another important performance measure is the dependence of the average user throughput versus the distance from the reference cell, as presented in Fig. 7.14. From the enlargement of Fig. 7.14 we can see that the drawback of the maximum C/I algorithm is that it discriminates users near the cell edge. Especially, the users at the cell edge (e.g. distances larger than 0.45 km) are starved for resources and have a rather low downlink throughput. The proportional fair algorithm has the best performance with regard to the user throughput versus distance as it makes the trade-off between the current channel conditions and previously experienced throughputs.

7.6

Semi-analytical Evaluation of Downlink Sub-carrier Allocation Algorithms

This section proposes an effective analytical framework to estimate the average sector throughput, user throughput, and user throughput versus distance for the resource allocation algorithms not relying on channel quality feedback that is Reuse

306

L. Jorguseski and R. Prasad Normalized User Throughput vs Distance; Pedestrian B; WiMAX 0.5

Re-use 1 Re-use 3 Soft re-use Maximum C/I Proportional Fair Soft partitioning

Normalized User Throughput [bit/s/Hz]

0.45 0.4 0.35 0.3

0.025

0.25

0.02

0.2

0.015 0.01

0.15

0.005

0.1

0 0.3

0.35

0.4

0.45

0.05 0

0

0.05

0.1

0.15

0.2 0.25 0.3 Distance [km]

0.35

0.4

0.45

0.5

Fig. 7.14 WiMAX user throughput versus distance for 30 users per sector

One, Reuse Three, and Soft reuse. The investigated WiMAX cellular system is presented in Fig. 7.4 with the following assumptions: l l l l

The cells are circular with radius R. Users are uniformly distributed over the whole target cell. No frequency selective multi-path fading.2 No shadowing effects in the propagation loss.

Let B ¼ {B1, B2, . . ., BQ} be a set of Resource Units (RUs) within the transmission bandwidth. Each RU consists of M¼28 for WiMAX. Average sector throughput of the qth RU is given by: p

Z 2 ZR Rq ¼

Rq ðr; yÞf ðr; yÞrdrdy p6

2

(7.11)

0

Note that in the simulation studies in Section 7.5 the frequency selective multi-path is taken into account in order to show the performance gain of Proportional Fair and Maximum C/I allocation algorithms. These algorithms utilize the frequency selectivity of the radio channel in the allocation of the resource units. However, the frequency selective multi-path is not considered in the semianalytical approach because the focus is first on the Reuse-One, Reuse-Three and Soft Reuse algorithms that do not utilize the frequency selectivity of the radio channel for the allocation decisions. This is left for further study.

7 Radio Resource Management in WiMAX Networks

307

f(r, y) is the probability density function of user’s location (r, y) Rq(r, y) is the achievable throughput per RU if allocated to a user on location (r, y) Because we have assumed that users are uniformly distributed, hence f(r, y) is a constant and has to satisfy the following condition: Zymax Zrmax f ðr; yÞrdrdy ¼ 1 ymin

(7.12)

rmin

After solving the double integration (7.12), we get: f ðr; yÞ ¼ 

2  2 2  rmin ðymax  ymin Þ rmax

(7.13)

Rq(r, y) is a function from effective SIR i.e. Rq(r, y) ¼ f(SIReffq(r, y)) [8]: ! M g ðr;yÞ 1 X  mb SIReffq ðr; yÞ ¼ b ln e (7.14) M m¼1 gm(r, y) is the Signal-to-Interference-and-Noise Ratio (SINR) of user at position (r, y) over sub-carrier m and given by:  gm ðr; yÞ ¼ PðmÞGq ðr; yÞ

 N RD N þ Np NSD =NST

(7.15)

We have assumed no frequency selective multi-path fading which means that P(m) ¼ 1 for all sub-carriers. For a UE located at a position (r, y) in the target cell, the average geometry of user at position (r, y) over qth RU Gq(r, y) is defined as the ratio of the total received power from the reference (desired) sector to the total aggregate power of all interfering sector (7.9):

Gq ðr; yÞ ¼

. PRU q;b;s Lb;s ðr; yÞ 19 P 3 P i¼1 j¼1

PRU q;i;j =Li;j ðr; yÞ

þ

with j 6¼ s if i ¼ b

(7.16)

RU Nth

The propagation loss consists of two factors: l

l

The distance based path loss. For WiMAX with 2.5 GHz operating frequency the distance based path loss can be calculated as [10] PLi(r, y) ¼ 130.18þ37.6Di(r, y). Here, Di(r, y) is the distance between user at position (r, y) and ith base station (km). When the distance between user and the base station is smaller than 0.03 km the path loss is set to 70 dB. The antenna gain has e beam pattern as defined in [8, 10]. For angle y (relative the boresight direction), the antenna has gain:

308

L. Jorguseski and R. Prasad

"



GðyÞ ¼ Gmax þ max 12

y

#

2

y3dB

; GFB

(7.17)

y 3dB is the 3dB beam width Gmax is the maximum antenna gain (boresight direction); Gmax = 16 dBi GFB is the front-to-back ratio; GFB = 25 dB In case of adaptive modulation and coding (AMC) the Rq(r, y) can be approximated by the sigmoid function as presented in Fig. 7.15: The sigmoid function has the following form: Rq ðr, yÞ ¼

Rmax 1 þ eabSIReff q ðr;yÞ

(7.18)

Rmax is the maximum transmission rate of AMCs per RU a, b are the fitting parameters (e.g. least square estimates fitting) The fitting parameters can be determined by, for example, least square estimate fitting of the sigmoid function to the link-level results Tables 3.1–3.4 in [9] for WiMAX. This is why the approach is labeled as semi-analytical. Replacing (7.18) in (7.12) and by using (7.14) and (7.15) we have average throughput of q-th RU per sector as follows: p

Z 2 ZR Rq ¼ p6 p 2

Z ZR p6

0

1þe

(7.19)

R  max 



¼

0

Rmax f ðr; yÞrdrdy abSIR effq ðr;yÞÞ ð 1þe

ab10log10 Gq ðr;yÞ



N NþNp

RD NSD =NST

 f ðr; yÞrdrdy

Throughput (Mbps)

16QAM

16QAM 1/2 QPSK 2/3 QPSK 1/3

S1

S2

S3

SIReff

Fig. 7.15 Approximation of the average user throughput per RU as a function of the effective SIR

7 Radio Resource Management in WiMAX Networks

309 R

Fig. 7.16 Division of the integration area according to the path loss calculation

A2 0.03 0

A1

The propagation loss is calculated based on two parts (smaller and larger than 0.03 km) as presented in Fig. 7.16. Hence, the average sector throughput of q-th RU is given by: p

Z 2 Z0:03 Rq ¼ A1 p6

0

Rmax f1 ðr; yÞrdrdy abSIR effq ðr;yÞÞ ð 1þe

p

Z 2 ZR þ A2 p6 0:03

Rmax f2 ðr; yÞrdrdy abSIR effq ðr;yÞÞ ð 1þe

(7.20)

A1, A2 are the ratios of the inner and the outer area to the whole sector area, respectively f1(r, y), f2(r, y) are the PDFs (7.12) of user’s location in the two integration parts. The average sector throughput is calculated as summation of the RU throughput as calculated in over the whole set Bq of available RUs. For calculating the average user throughput given the distance from the reference base station we can divide the covered cell area in concentric rings. The inner circle of the l-th ring is on distance rl while the outer circle in on distance rl+1, where the ring width is equal to rl + 1  rl as Fig. 7.17 and depends on the desired granularity for the throughput versus distance calculation. Nu is the number of users per sector. 0 1 Zymax Zrlþ1 BQ 1 X B C Rðrl Þ ¼ Rq ðr; yÞf ðr; yÞrdrdyA @ Nu q¼1 0

rl

ymin

Z BQ 1 X B ¼ @ Nu q¼1

p 2

p6

1

Zrlþ1 rl

 1þe

R  max

ab10log10 Gq ðr;yÞð

Þ

RD N NþNp NSD =NST

 f ðr; yÞrdrdyC A (7.21)

310

L. Jorguseski and R. Prasad

Fig. 7.17 The illustration of the user throughput versus distance calculation

R r1+1 r1

0

7.6.1

Comparison of Semi-analytical and Simulation Results

In order to compare the performance estimation of the semi-analytical approach with the simulation results the same simulation set-up is used as in Section 7.5 except for the following adjustments: l

l

The effect of shadowing and frequency selective multi-path fading were excluded from the simulations. Only the Reuse One, Reuse Three, and Soft Reuse algorithms were simulated for the WiMAX system that is the algorithms that do not rely on channel quality feedback information.

Due to the three-sector base stations and the corresponding antenna pattern in the investigated OFDMA cellular system the contour of the integrations for calculating the average sector and user throughput versus distance according has to be determined. For this purpose coverage only simulations were used to determine the angles and distances where the coverage is provided from the reference base station as illustrated in Fig. 7.18. These coverage only simulations were needed also to determine the integration contours for the Soft Reuse algorithm, depending on the selected geometry threshold of 8 dB (see Section 7.4.3). The goal is to determine the angles and distances where users are classified as “close” and “edge” as illustrated in Fig. 7.19 The semi-analytical and simulation results in term of average sector throughput and user throughput versus distance are presented in Table 7.3 and Fig. 7.20. Note that the relative errors are with regard to simulation results without the shadowing and frequency selective multi-path fading effects. From Table 7.3, we can see that the semi-analytical results of average sector throughput are approximately equal to the simulation results with relative error up to 3.7%. Additionally, the semi-analytical curves for average user throughput versus distance are quite close to simulation curves in Fig. 7.20. The existing relative errors as well as the slight deviation of semi-analytical curves and simulation curves can be explained by the accuracy of least squares fitting method used to find the

7 Radio Resource Management in WiMAX Networks Fig. 7.18 The illustration of investigated parts of the sector

pi/2 rad

311

1.51 rad 1.45 rad 1.33 rad 0.64 rad 0.6 rad 0.76 rad

0.46 rad 0.4 rad 0.28 rad

– 0.28 rad – 0.40 rad 0.44 km – 0.46 rad 0.46 km 0.47 km – pi/6 rad 0.49 km

p/2

Fig. 7.19 The classification of “close” users and “edge” users

p/3 “edge” users 0.25 km

0

0.3 km “close” users

– p/6

Table 7.3 Semi-analytical and simulation results of normalized sector throughput Allocation algorithm Semi-analytical (bit/s/Hz) Simulation (bit/s/Hz) Relative errors (%) Reuse one 0.5739 0.5688 0.90 Reuse three 0.5127 0.5139 0.23 Soft reuse 0.9533 0.9900 3.7

312

L. Jorguseski and R. Prasad Average User Throughput vs Distance; WiMAX

Normalized User Throughput [bit/s/Hz]

0.12

Reuse One - Simulation Reuse One - Semi-analytical Reuse Three - Simulation Reuse Three - Semi-analytical Soft Reuse - Simulation Soft Reuse - Semi-analytical

0.1

0.08

0.06

0.04

0.02

0

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

Distance [km]

Fig. 7.20 User throughput versus distance of semi-analytical and simulation results, 30 users/ sector, WiMAX

sigmoid function for the user throughput and effective SIR per RU and possible errors in the determination of the integration contours in Figs. 7.18 and 7.19 due to the finite granularity of the location estimations. In summary, the semi-analytical results are reliable when compared with simulation results without shadowing and frequency selective multi-path fading for the sub-carrier allocation algorithms Reuse One, Reuse Three, and Soft Reuse. The semi-analytical approach is valuable because it saves excessive time for calculating the average sector/user throughput and user throughput versus distance in OFDMA cellular systems. For example, for each point on the graphs presented in this section the semi-analytical approach needed few minutes compared with the simulation time (on a standard PC) of more than one day for 30 users/sector.

7.7

Conclusions and Recommendations

In this chapter four different sub-carrier allocation algorithms with low complexity are evaluated for WiMAX cellular systems with respect to the average sector/user throughput, and the average user throughput versus distance from the reference base station. The simulation analysis presented in this chapter has shown that the maximum C/I based allocation algorithm has the best performance with regard to the average sector and user throughput. However, it is highly unfair for the users on locations further

7 Radio Resource Management in WiMAX Networks

313

away (e.g. roughly 0.3 km) from the reference base station because they are “starving” for resources resulting in the lowest user throughput when compared with the other sub-carrier allocation algorithms. The proportional fair sub-carrier allocation algorithm mitigates this unfairness and has significantly more even distribution of the downlink throughput for different distances from the reference base station when compared to the maximum C/I algorithm. It clearly outperforms all other algorithms for distances roughly larger than 0.25 km from the reference base station. For example, at the sector edge the user throughput for the proportional fair algorithm is roughly doubled when compared to the Reuse One (i.e. the reference case). The soft-reuse allocation algorithms have similar sector and user throughput when compared to the proportional fair allocation algorithm. However, the user throughput at cell edge is lower when compared with the proportional fair algorithm. The Reuse One and Reuse Three sub-carrier allocation algorithms have the worse downlink throughput performance when compared to the other sub-carrier allocation algorithms. This is the result of purely random allocation of the RUs to different users and, in case of Reuse Three, using only a portion of the available bandwidth. From this relative throughput performance comparison of the different subcarrier allocation algorithms it can be concluded that proportional fair sub-carrier allocation algorithm is recommended for an OFDMA cellular system as it has relatively high sector/user throughput and has a fair distribution of the user throughput versus the distance from the reference base station. The semi-analytical approach for the downlink throughput evaluation of Reuse One, Reuse Three, and Soft Reuse sub-carrier allocation algorithms has shown reasonable accuracy when compared with simulation results without the effects of shadowing and frequency selective multi-path fading. The relative error for the average sector/user throughput was up to 3.7% for WiMAX. The low relative error and drastically reduced processing time (minutes versus days) make this semianalytical approach very useful for practical use. It is recommended to evaluate in more detail the power allocation per RU for the Soft Reuse algorithm and the division of the total amount of RUs into a block of resources available either for the “close” or “edge” users. The more detailed investigation should include a “dynamic” allocation of the number of RUs with higher power (intended for the “edge” users) depending on the amount of “edge” users in the sector. Regarding the semi-analytical approach it is recommended to include the effect of shadowing and frequency selective multi-path fading in the model and investigate the accuracy of the approach by comparing the semi-analytical results with simulations of the maximum C/I and proportional fair sub-carrier allocation algorithms.

References 1. J.G. Proakis, Digital Communications – Second Edition (McGraw-Hill, New York, 1989). ISBN 0-07-050937-9 2. C. Lee, L.H.C. Lee, Convolutional Coding: Fundamentals and Applications (Artech House, Boston, MA, 1997)

314

L. Jorguseski and R. Prasad

3. L. Hanzo, T.H. Liew, B.L. Yeap, Turbo Coding, Turbo Equalisation and Space-time Coding for Transmission Over Fading Channels (Wiley, New York, July 2002) 4. WiMAX Forum, Mobile WiMAX – Part I: A Technical Overview and Performance Evaluation (Aug 2006) 5. WiMAX Forum, Mobile WiMAX – Part II: A Comparative Analysis (Aug 2006) 6. L. Jorguseski, R. Prasad, Downlink resource allocation in beyond 3G OFDMA cellular systems, in 18th Annual IEEE International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC’07), Athens, Greece, 3–7 September 2007 7. J.-G. Choi, S. Bahk, Cell throughput analysis of the proportional fair scheduling policy, in Proceedings of NETWORKING 2004, Athens, Greece, May 2004 8. 3GPP TR 25.892, Feasibility Study for Orthogonal Frequency Division Multiplexing (OFDM) for UTRAN enhancements, V 6.0.0 (June 2004) 9. IEEE 802.16 Broadband Wireless Access Working Group. Draft IEEE 802.16m Evaluation Methodology (14 Dec 2007) 10. WiMAX forum, WiMAX system evaluation methodology, Version 2.0 (15 Dec 2007)

Chapter 8

Radio and Network Planning Fernando J. Velez, Pedro Sebastia˜o, Rui Costa, Daniel Robalo, Cla´udio Comissa´rio, Anto´nio Rodrigues, and A. Hamid Aghvami

Abstract This chapter starts by presenting the Stanford University Interim (SUI) and modified Friis propagation models. Although the SUI model is being recommended for WiMAX, the comparison between the model and experimental results show that, in our environment, at 3.5 GHz, the modified Friis model with g = 3 fits better the measurement values. From the analyses of the signal-to-noise-plusinterference ratio, SNIR, interference-to-noise ratio and reuse pattern, it is found that both noise and interference present a strong limitation to the performance of fixed WiMAX, mainly for higher order modulation and coding schemes (MCSs). In general terms, the use of sectorization in fixed WiMAX enables to reduce the reuse pattern while considering sub-channelisation allows for improvement on the coverage. The reduction of the reuse pattern directly corresponds to an increase in the system capacity but the improvement in the coverage range (through subchannelisation) can also allow for an improvement in UL system capacity, as adaptive MCS are used. Two different approaches are considered for graphical cellular planning, and the district of Covilha˜ is considered as a case study. On the one hand, one considered a GIS based WiMAX planning tool conceived by considering coverage issues, frequency reuse, and the impact of the different classes of service. On the other, WinpropTM is used as it distinguishes among different MCS in the graphical presentation of the results. Both tools consider the information coming from the digital terrain profile. The GIS functionalities allow for appropriately adjusting azimuth and tilt of antennas. This cellular planning exercises confirm the results of theoretical analysis, where different crowns are achieved for the coverage with each MCS (corresponding to a given range of values for SNIR), for the maximum physical throughput, and for the “best server” cells. The frequency radio resources should be considered as the most valuable resource

F.J. Velez (*) Instituto de Telecomunicac¸o˜es-DEM, Universidade da Beira Interior, Calc¸ada Fonte do Lameiro, 6201-001, Covilha˜, Portugal e-mail: [email protected]

R. Prasad and F.J. Velez, WiMAX Networks, DOI 10.1007/978-90-481-8752-2_8, # Springer ScienceþBusiness Media B.V. 2010

315

316

F.J. Velez et al.

during the planning of wireless broadband access networks. As a rule, spectral efficiency needs to be optimized by using several advanced techniques, corresponding to an optimization from the cost-benefit point of view.

8.1

Introduction

The present context of frequency spectrum management worldwide, as well as the need for making broadband access flexible for users, creates opportunities for the entrance of new operators and offer diversification of innovative access technologies. As Worldwide Interoperability for Microwave Access (WiMAX) enables the support of mobile broadband Internet services in outdoor (and even in indoor, for the lower frequency bands) with high coverage ranges and user mobility support. It allows the exchange of truly wide and broadband multimedia content, and support simultaneously all-Internet Protocol (IP) voice, data, streaming, image and video multi-rate communications. The goal of cellular coverage is to provide access to mobile users in a given region, called cell, while guaranteeing the quality of the received signal in both directions, Uplink (UL) and Downlink (DL), even for the users at the cell edge. As resources, for example, frequency channels, need to be reused in different geographical zones (but not in close proximity), the impact of interference among co-channel cells needs to be evaluated also in both directions. In WiMAX, as in Wideband Code Division Multiple Access (WCDMA) and High Speed Downlink Packet Access (HSDPA), the ideal situation would be to reuse the channels in every cell, that is, to deploy systems with a frequency reuse pattern, K, equal to one, which would be achieved by means of Pseudo Random Mapping (PRM) of sub-carriers where Orthogonal Frequency Division Multiple Access (OFDMA) is used. However, due to heavy interference in frequency reuse deployment, users at cell edge may suffer from low connection quality since these improvements may not be available in some versions of the standard, or may simply reduce but not eliminate the interference. In the context of WiMAX planning, research on the variation of the carrier-tonoise-plus-interference ratio (CNIR) with different system parameters is therefore of paramount importance. As there are limitations in both links, UL and DL, techniques such as sub-channelisation need to be applied to reduce the impact of the noise on the link performance. However, only mobile WiMAX will allow for sub-channelisation in the DL while fixed WiMAX only allows for it in the UL, and may cause a degradation of performance (mainly owing to the extra noise caused by the largest bandwidth). For cellular planning purposes, the UL and DL CNIRs from/at the wireless Subscriber Station (SS) are very important parameters. From a detailed analysis of its variation with the coverage and reuse distances for different modulation and coding schemes (MCS), an evaluation of the possible reuse patterns can be performed.

8 Radio and Network Planning

317

If Frequency Division Duplexing (FDD) is used, in fixed WiMAX, worst-case situations occur in the DL when the Base Station (BS) of the central cell transmits to the most distant SS, located at the cell edge, whilst receiving interference from the BS of the six co-channel cells of each ring of interference (Fig. 8.1). In the UL, the worst-case situation occurs when the SS is transmitting to the BS from the cell boundary while interfering mobiles are at the interfering cells edge (in the region closest to the central cell) (Fig. 8.2). When sectorization is considered the number of interfering cells is decreased, and system capacity increases. Usage scenarios will be enabled by using innovative terminals, similar to PDAs or Tablet PCs, which will combine voice with other type of services, including image and video. One example can be the communication of real-time image from an actual fire site to the fire department. In Summer time, in the south of Europe, simultaneous fires in forests are a persistent calamity, and authorities lack access to real-time fire information in order to coordinate fire brigades. Another good example is the surveillance of commercial streets by using real-time video. For demonstration, a network was deployed in the city of Covilha˜, Portugal, by using IEEE 802.16-2004 BreezeMAX Alvarion equipment with 3.5 MHz channels at 3.5 GHz [1], as indicated in Table 8.1 (where supported data rates are presented). Later, by using the IEEE 802.16e standard [2], it will be possible to support true mobility. Different received power levels correspond to different net physical (PHY) bit rates, and to different modulation and coding schemes (MCSs), as shown in Table 8.2 (for the Alvarion BreezeMAX 3.5 MHz equipment). This Point-to-Multipoint (PtM) network is the basic tool for our research inbroadband mobile access, mobile IP, and always best connected WiMAX scenarios, including the possibility of performing extensive field trials in the FDD mode. In the initial phase of WiMAX deployment, the operators experience is limited, and measurement based cellular planning procedures, extracted from the experimental

D+R D

D–R

Fig. 8.1 Co-channel interference in the worst-case for the DL

D+R

R D D–R

318

F.J. Velez et al.

Fig. 8.2 Co-channel interference in the worst-case for the UL

D– R D–R

D –R

R

D–R

Table 8.1 Total data rates for IEEE 802.16-2004

Bandwidth (MHz) 3.5 5.0 7.0 10.0 20.0

Table 8.2 Correspondence among modulations, sensitivity, and net PHY rate

Modulation & coding BPSK 1/2 BPSK 3/4 QPSK 1/2 QPSK 3/4 QAM 16 1/2 QAM 16 3/4 QAM 64 2/3 QAM 64 3/4

QPSK 3.3 4.6 6.5 9.3 18.7

D–R D–R

Data rate (Mbps) 16-QAM 6.5 9.3 13.1 18.7 37.5

Net PHY bit rate (Mbps) 1.41 2.12 2.82 4.23 5.64 8.47 11.29 12.27

64-QAM 9.8 13.9 19.6 28.0 56.2

Sensitivity (dBm) 100 98 97 94 91 88 83 82

networks and prototypes, are fundamental at for the validation of the proposed planning algorithms. A Geographic Information Systems (GIS) based planning tool was developed for the purpose of helping in the process of design WIMAX coverage and frequency reuse. It includes the propagation models, the co-channel interference computation aspects, as well as the services and applications details. However, it does not fully includes a detailed analysis of the consequences of using different MCS, and a

8 Radio and Network Planning

319

comparison with a commercially available planning tool, as WinpropTM [3] was justified. This specific functionality of WinpropTM allows for helping to verify the impact of the improvement strategies, as sub-channelisation and sectorization. Different operation “modes” and the information coming from the digital terrain profiles will be considered in the context of specific cellular planning exercises. The structure of this Chapter is as follows. Section 8.2 presents the point-tomulti-point propagation measurements and models, namely the modified Friis and the Stanford University Interim (SUI) models. Section 8.3 addresses aspects of cellular planning. It starts by discussing the limitations of a simplified analysis, and then a comparison of the results for the interference-to-noise ratio and the achievable reuse pattern is performed. The variation of the CNIR with co-channel reuse factor is used as a basis to compute the supported PHY throughput for different cases (by considering the absence and presence of sub-channelisation and of sectorization). Section 8.4 presents results obtained by using two wireless planning tool platforms, one developed by ourselves while the other one is commercially available. The framework and scenarios are analysed, the functionalities and potentials of the tools are discussed, and planning results are given for different environments as well. Finally, in Section 8.5 conclusions are made as well as suggestions for further work.

8.2 8.2.1

Propagation Models SUI Versus Modified Friis Model

In Non Line-of-Sight (NLoS) channel conditions, signals may undergo scattering, diffraction, polarization changes and reflection impairments, which affect their level and phase at the receiver. Usually these impairments are not important if there is Line-of-Sight (LoS) between the transmitter and the receiver. For outdoor environments, obstacles, such as building materials, foliage and clutter, also contribute to increase path loss [4, 5], and the SUI outdoor propagation model is especially relevant [4]. It will be considered together with the modified Friis model. Over the years, various models have been developed to characterize Radio Frequency (RF) environments and allow for the prediction of the RF signal strengths. These empirical models are used to predict large-scale coverage for radio communications systems in cellular applications and provide estimates for path loss (PL) by considering the distance between the transmitter and receiver, terrain factors, antenna height, and cellular frequencies. Nevertheless, according to [4] none of these approaches address the needs of broadband fixed wireless adequately. To overcome this limitation, AT&T developed an empirical wireless model that has been validated against deployed fixed wireless systems which yielded results comparable to other models and experiments. This model was the

320

F.J. Velez et al.

basis of an industry-accepted model, and is being used by the standardization bodies, such as in the IEEE 802.16. The adoption of the AT&T wireless PL model by IEEE is referred as “Channel Models for Fixed Wireless Applications” in [5]. The AT&T wireless PL model includes, as parameters, antenna heights, carrier frequency and types of terrain [4]. Apart from the modified Friis model (where the propagation exponent g instead of taking a value of 2 is replaced by a different empirical value), one possible solution for the Wireless Planning Tool (WPT) is therefore the SUI model, which is an extension of the earlier work of AT&T propagation model [4]. The SUI model uses three basic terrain types: l l l

Category A – Hilly/moderate-to-heavy tree density Category B – Hilly/light tree density or flat/moderate-to-heavy tree density Category C – Flat/light tree density

These terrain categories provide a simple method to estimate more accurately the PL of the RF channel in an NLoS situation. Being statistical in nature, the model is able to represent the range of PLs experienced within an actual RF link. SUI models were explored for the design, development, and testing of WiMAX links in six different scenarios, SUI-1 to SUI-6 [6]. By using these propagation models (both modified Friis and SUI ones), it is then possible to predict more accurately the coverage probability achieved within a base station site sector. These models do not replace detailed site planning (and site surveying), but can provide an estimate before actual planning. Besides, it is very important to perform RF planning activities to adequately evaluate specific environment factors, co-channel interference, actual clutter and terrain effects. This model allows several frequencies and SSs heights. The path loss is given (in dB) by [7] PLðdÞ ¼ PLðd0 Þ þ 10  g  logðd=d0 Þ þ Xf þ Xh þ S;

(8.1)

with the following parameters PLðd0 Þ½dB ¼ 20  logð4pd0 =lÞ;

(8.2)

g ¼ a  b  hb þ c=hb ;

(8.3)

and

where d is the distance between the BS and a given point, in meter, d0 ¼ 100 m, l ¼ c=f is the wave length, c ¼ 3  108 ms1 , f is the carrier frequency, hb is the BS height above ground, in meter (10 < hb < 80 m), and a, b, and c are parameters which are chosen according to three environments, represented by A, B or C, Table 8.3. The terms Xf and Xh are correction factors for frequency and SS antenna height above the ground, respectively.

8 Radio and Network Planning Table 8.3 Values for the parameters in the SUI model

321 Model constant a b c

A 4.6 0.0075 12.6

Terrain type B 4.0 0.0065 17.1

C 3.6 0.0050 20

These correction factors are defined as
 Xf ¼ 6:0  log f=2000 ;

(8.4)

and
 8 < 10:8  log hm= ; for terrain types A and B 2:0
 ; Xh ¼ : 20:0  log hm= 2:0 ; for terrain type C

(8.5)

where f is the carrier frequency, in MHz, and hm is the receiver height above the ground, in meter. The term S is a lognormal-distributed random variable with zero mean and standard deviation sS, with typical values from 8.2 to 10.6 dB, depending on the type of terrain [7]. This term takes shadow fading originated by trees and structures into account [7]. The average received power is given byPRðdÞ ¼ PE  PLðdÞ, where PE is the average transmission power and PLðdÞ is the average path loss (attenuation factor). In the context of other commercially available cellular planning tools, models with high complexity can be used, namely, the NLoS dominant ray path loss one [7, 8]. However, in order to obtain the most efficient cellular planning tools it is important to compare these models with the simplest ones, for example, the modified Friis and SUI ones, and validate them against experimental results.

8.2.2

Experimental Results

Figure 8.3 presents some of the PtM Alvarion WiMAX equipment used in the experimental setup installed on the roof top of the Health Sciences Faculty (HSF) of University of Beira Interior, Covilha˜, near the Hospital. They are a BreezeMAX 3000 OFDM micro BS and self-installable Alvarion BreezeMAX Customer Premise Equipments (CPEs) operating at 3.5 GHz, an omnidirectional antenna, and the Outdoor Unit (ODU). The appropriate Ethernet and RF cables were also used, as well as a Global Positioning System (GPS) device, a 12–240 V power inverter (to feed the CPE), and a portable PC, used as a terminal and running a File Transfer

322

F.J. Velez et al.

Fig. 8.3 IEEE 802.16-2004 PtM equipment operating at the 3.5 GHz band

30 25

~550 m 20 15 10

500 m

5

Fig. 8.4 LoS regions and measurements of SNR[dB] for the DL at 3.5 GHz

Protocol (FTP) application, the BreezeCONFIG software, and a tool for acquiring GPS positions. In the HSF backhaul network, there was an FTP server, and a Dynamic Host Configuration Protocol (DHCP) server, for automatically assigning an IP address to the users’ CPEs. The BS antenna gain is 10 dBi (360 azimuth, 8 elevation, vertical polarization). The SS antenna is a beam switching array of six 9 dBi, 60 antennas, integrated into the CPE. Field trials were performed in the suburban area of Covilha˜, in a zone with approximately 2.80  1.55 km2, and initial results for signal-to-noise ratio (SNR) and the throughput were obtained at the SSs (or CPEs) that roam around the suburban area surrounding the HSF. The dynamic range of the CPE was found to be adequate. There is a direct correspondence between the SNR values from Fig. 8.4, the MCSs, and the achieved data rate, for example, 6 Mbps for 16-QAM. The BreezeMAX duplexing frequency range is 3,499.5–3,553.5 MHz and 3,550–3,600 MHz for downlink (DL), and 3,399.5–3,453.5 MHz, and 3,450–3,500 MHz for uplink (UL) [9]. In these particular field tests, our ODU was operating at 3,551.75 MHz (DL), and 3,451.75 MHz (UL), and with a transmitter power of 28 dBm. NLoS regions, obtained with ArcGIS, are also represented

8 Radio and Network Planning

323

80 70 60

C/N [dB]

50

C/N= – 8.259log(d ) + 25.417

40 30 20 10 0 –10 –20

0

0.2

0.4

0.6

0.8 d [km]

1

1.2

1.4

Friis (g = 2)

Friis g = 3)

Friis (g = 4)

Measured

SUI (type A)

SUI (type B)

SUI (type C)

Trendline

1.6

Fig. 8.5 Trend curve for the measured SNR in the locations around HSF: comparison between the SUI and the modified Friis models

in the background, in green, in Fig. 8.4. At distances far from the base station, the experimental regions with reasonable signal quality coincide basically with the LoS regions. However, for distances up to 550 m even NLoS areas near the boundary between LoS and NLoS areas can be covered. Another important issue is the comparison between the results for the curve fitting of SNR and the curves obtained from the application of the propagation models (SUI model and modified Friis equation), as depicted in Fig. 8.5, for distances up to 1.6 km. Within the modified Friis model different propagation environments are modelled by different propagation exponents, g, which vary from g ¼ 2, corresponding to free space conditions, for example, rural areas, to g ¼ 3, in urban areas (no shadowing), and g ¼ 4, in shadowed urban areas [10]. For the SUI model, one considers hm ¼ 2 m, hb ¼ 13.3 m and sS ¼ 8.8 dB. For both models, a total antenna gain (transmitter plus receiver) of 19 dBi, a transmitter power of 28 dBm, a bandwidth of 3.5 MHz, and a noise factor of 3 dB were assumed. To understand the results, an analysis of the terrain profile around the BS location at HSF is required. While at Southeast (SE) of the BS the terrain is flat, at Northwest (NW) it is continuously hilly, with increasing altitudes. The first coverage circular zone, with a radius of 550 m from the BS, is part of the suburban region of Covilha˜, with a moderate building density. The second coverage zone (or crown), for distances larger than 550 m, corresponds to a rural area (and the small airport) at Southeast, and to a dense urban area at NW. However, this dense urban area corresponds to zones predominantly in LoS with the BS as the terrain height continuously increases from the BS to this zone. At SE, the zones up to a distance of 550 m are predominantly in non-LoS owing to the shadowing effect of the HSF roof. However, at NW of the BS, this crown is

324

F.J. Velez et al. 38

C/N [dB]

32

C/N = – 32.72log(d) + 15.86

26

20

14 0.25

0.3

0.35

0.4

0.45

0.5

d [km] Friis ( g = 3) SUI-B

Measured SUI-C

SUI-A Trendline

Fig. 8.6 Analysis of the measured SNR for distances in the interval [275, 475] m: comparison between the SUI and the modified Friis models

predominantly in LoS (as for higher distances), as the terrain height increases. For distances larger than 550 m measurements were taken mostly at locations with LoS coverage. A curve fitting approach was used to interpret the measurement results from Fig. 8.5 (in dB). The coefficient 8.259 for log(d) corresponds to a very low power decay rate. If the modified Friis equation is considered, the experimental results take values similar with the curve for g ¼ 4 for distance in the interval (60, 150 m), similar to the ones from the curve for g ¼ 3 for distances in the interval (300, 500 m). Then, for larger coverage distances, values correspond to modified Friis equation propagation exponents between 2 and 3. It seems that the results consecutively correspond to (1) shadowed urban areas, (2) urban areas (with no shadowing), and (3) approximately free space, which is partially true if we consider the actual terrain. Despite the fact that the use of the SUI model is being recommended for WiMAX, the comparison of the experimental results with the SUI model is harder since they only coincide for distances in the interval (275, 475 m). In order to clarify this issue, Fig. 8.6 presents a partial analysis of the results for C/N in this interval, where a different trendline is obtained, and only the modified Friis equation curve for g ¼ 3 is considered for comparison, as it approximately follows the experimental trendline for the results with a difference of circa 2.5 dB (which can be caused by the penetration loss due to the placement of the SS antenna in the rear seat of the car during the measurements). If this difference of 2.5 dB was subtracted from results for modified Friis equation with g ¼ 3, the curves would be almost coincident. This was confirmed by the reduction of the Mean Square Error (MSE) from 4.917 to 0.113.

8 Radio and Network Planning

325

Without this correction factor, while the distance to the BS increases the trend line consecutively crosses the curves for SUI-A, SUI-B, and SUI-C, that is, the curves for (a) hilly/moderate-to-heavy tree density, (b) hilly/light tree density or flat/moderate-to-heavy tree density, and (c) flat/light tree density, respectively. The experimental results cross the three curves of the SUI model but a good correspondence was only found for the SUI-B and SUI-C, as the MSE are 1.564 and 1.617, respectively. For the SUI-A curve, the MSE was 4.657. However, if we apply the correction factor of 2.5 dB (as for the modified Friis equation) the results for the SUI-A and SUI-B models would be much worst (MSE of 19.227 and 9.247, respectively). Only the curve for the SUI-C would still offer an almost acceptable correspondence (with an MSE of 4.808). Results from field trials show that, at 3.5 GHz, the propagation at distances higher than 550 m is mainly in LoS; hence, a need is identified of using lower frequency bands to achieve the objective of having appropriate NLoS propagation within WiMAX networks. Another important result at this band is the appropriateness of the modified Friis propagation model for this band. Despite the use of the SUI model is recommended for WiMAX, in these experiments at 3.5 GHz, for distances in the interval (275, 475 m), the experimental results fit quite well with the modified Friis equation with #  3, which corresponds to urban areas (with no shadowing), although the SUI-C model can also be a solution.

8.3 8.3.1

Cellular Planning Limitations of a Simplified Analysis

In order to achieve an efficient use of the radio frequency spectrum, it is important to choose a frequency reuse scheme that leads to coverage guarantee, and improved system capacity whilst minimising the interference. If FDD is used, an analytical approach may be followed in fixed WiMAX (IEEE 802.16-2004) to solve coverage and frequency reuse problems. Traditionally, several text books decouple the analysis of the carrier-to-noise ratio aspects from the analysis of the carrier-to-interference ones. This way, interference is not considered in the coverage problem while noise is not considered when addressing the frequency reuse. On the one hand, not considering the interference in the coverage problem means that the maximum coverage distance is obtained without allowing for extra margin for the interference, which is a critical limitation for the dimensioning process of a cellular system. On the other hand, the absence of noise in the interference analysis is unrealistic, as noise is always present in communications. As there would be limitations arising from an independent analysis, cellular planning has to consider simultaneously carrier-to-noise and carrier-to-interference constraints. Improvement techniques as sub-channelisation and sectorization, will be addressed to improve coverage and avoid interference.

326

8.3.2

F.J. Velez et al.

Signal-to-Noise-Plus-Interference Ratio

This Section addresses aspects related to the analysis of the cell coverage distance (or cell radius), R, the co-channel reuse factor, rcc, the ratio between the reuse and the coverage distances, the reuse pattern, K, for several levels of IEEE 802.16 modulation and coding schemes. The conclusions to be extracted facilitates an adequate choice of the reuse pattern and an efficient frequency planning. If one considers the interference-to-noise ratio, defined by M ¼ I=N ;

(8.6)

and the equation for the carrier-to-noise-plus-interference ratio (CNIR) to be used in the dimensioning process is C ¼ ðN þ I Þ

  C ; N min

(8.7)

Equation (8.7) can therefore be re-written in the two following ways   C ð1 þ MÞ; N min

(8.8)

    C 1 þ M1 : N min

(8.9)

C ¼ N and C ¼ I

In (8.7) one is using the model for CNIR from [11] while assuming that the weights for noise and interference are the same. From Eq. (8.8) one obtain the following equation for the interference-to-noise ratio MðRÞ ¼

ðCðRÞ=N Þ  1; ðC=N Þmin

(8.10)

where C(R) ¼ PR(R) is computed by applying the modified Friis formula with g ¼ 3, the hypothesis followed in this Section for urban environments. This value for the propagation exponent results from the experimental results in Covilha˜, Portugal. Values of M(R) are proportional to the interference still tolerable for a given coverage distance R while (still) agreeing with the quality requirements for a given MCS.

8 Radio and Network Planning

327

With a hexagonal cell topology, in the DL (Fig. 8.1) as the distance associated with interference is D, that is, the reuse distance itself, the carrier-to-interference ratio can be given by C 1 rcc g ¼ ;  I 2ðrcc þ 1Þg þ 2rcc g þ 2ðrcc  1Þg 6

(8.11)

where rcc is the co-channel reuse factor, given by rcc ¼ D=R:

(8.12)

The approximate expression in (8.11) is very useful in practice. For the UL (Fig. 8.2) the carrier-to-interference ratio is given by C ðrcc  1Þg ¼ : I 6

(8.13)

By replacing (8.11) into (8.9), it is therefore possible to obtain the following equation for the reuse factor in the DL rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi g C 1 : rcc ¼ 6  ð1 þ M Þ  =N min

(8.14)

For the UL, in turn, by replacing (8.13) into (8.9), one obtain the following equation rcc ¼

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi g þ 1: 6  ð1 þ M1 Þ  C=N min

(8.15)

It is worthwhile to note that, for hexagonal reuse geometries, the reuse pattern is given by K¼

rcc 2 : 3

(8.16)

As a horizontal asymptote arises in the analysis of the curves of rcc as a function of the coverage distance, R, it is important to present the mathematical details associated to it. To compute the horizontal asymptote in the chart of rcc(R), one has to consider that R ! 0. From (8.10), if R ! 0 then M ! +1, and M1 ! 0. On the one hand, for the DL, in the limit, one obtains rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi g C : lim rcc ¼ 6  =N

R!0

min

(8.17)

328

F.J. Velez et al.

On the other, for the UL, also in the limit, one obtains rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi g þ 1: lim rcc ¼ 6  C=N R!0

min

(8.18)

By considering (8.16), it is straightforward to conclude that, for each value for the propagation exponent, the reuse pattern, K, only depends on the MCS through the value of the corresponding minimum carrier-to-noise ratio, as well as on the cellular interference geometry, either UL or DL. While the asymptotic reuse factor is associated with the upper bound for system capacity, the maximum coverage distance is associated with the carrier-to-interference-plus-noise ratio at the cell boundary when the interference is null. Since the interference-to-noise ratio, M, represents the interference that can still be tolerated for a given R, in the limit, the maximum coverage distance for which no extra interference is tolerated is obtained when I(R) ! 0, that is, when M(R) ! 0 (meaning that M[dB] ! 1). Hence, the vertical asymptotes for the M(R) and rcc(R) charts is Rasymptote ¼ RM!0 :

(8.19)

It is obtained by solving the following equation MðRÞ ¼

ðCðRÞ=N Þ  1 ¼ 0; ðC=N Þmin

(8.20)

or, in a simplified way
 CðRÞ= ¼ C= N N min :

(8.21)

By comparing this equation (valid only when M ! 0) with Equation (8.8), one concludes that only considering the carrier-to-noise ratio to determine the coverage distance, R, is inadequate in systems where interference is relevant, as Equation (8.21) corresponds to a null interference-to-noise ratio, M. If a cellular system was dimensioned this way there would not be an extra margin for interference, represented by M ¼ I/N. Finally, it is a worth noting that, for a given propagation exponent, the maximum coverage distance corresponding to the vertical asymptote, Rasymptote, depends not only on the MCS but also on the noise power, N. This is the reason why the reduction of the noise power through sub-channelisation, that is, through the reduction of the RF bandwidth, is so important.

8.3.3

Interference-to-Noise Ratio and Reuse Pattern

By using (8.10), one obtains the chart for the interference-to-noise ratio without sub-channelisation from Fig. 8.7, which is valid both for the UL and DL. The considered propagation exponent for the modified Friis model is g ¼ 3.

8 Radio and Network Planning

329

50

BPSK3/4 QPSK1/2

40

QPSK3/4

30

16QAM1/2

M [dB]

16QAM3/4

20

64QAM2/3 64QAM3/4

10 0 –10 –20 0

1000

2000

3000 R [m]

4000

5000

6000

Fig. 8.7 Interference-to-noise ratio without sub-channelisation

Table 8.4 Values for the vertical asymptote without sub-channelisation

Level 1 2 3 4 5 6 7 8

MCS BPSK 1/2 BPSK 3/4 QPSK 1/2 QPSK 3/4 16-QAM 1/2 16-QAM 3/4 64-QAM 2/3 64-QAM 3/4

Rasymptote (m) 5,814.86 4,911.41 4,548.55 3,783.32 2,936.79 2,368.86 1,638.85 1,494.65

The modified Friis propagation model with g ¼ 3, transmitter power Pt ¼ 2 dBW, and transmitter and receiver antenna gains Gt ¼ 17 dBi and Gr ¼ 9 dBi, respectively, were assumed. Note that transmitter antenna gain is 7 dB higher than the value considered in Section 8.2.2. The radio frequency bandwidth, the noise figure and the frequency were brf ¼ 3.5 MHz, Nf ¼ 3 [dB], and f ¼ 3.5 GHz, respectively. Table 8.4 presents the corresponding values for the vertical asymptote without sub-channelisation, Rasymptote. It is observed a relevant decrease of the values for the vertical asymptote (maximum coverage distance) as the MCS level increases, which is compatible with the lowest values for SNRmin. By applying (8.14) and (8.16) to the DL one obtains the charts for rcc(R) and K (R) from Figs. 8.8 and 8.9, respectively. By using a reuse pattern K ¼ 7 it is possible to use a maximum MCS level of 4, that is, QPSK , for R b 2.7 km, and a maximum MCS level of 5, that is, 16-QAM ½, for coverage distances lower than 1.2 km. Figure 8.10 presents the results for the UL, which are worst since they correspond to higher values for the reuse pattern. In the UL without sub-channelisation, for a reuse pattern K ¼ 7, only a low order MCS is achievable, that is, QPSK , up to a coverage distance of 2 km. It is not possible to use 16-QAM ½, as it was in the DL.

330

F.J. Velez et al. 14

BPSK1/2 BPSK3/4 QPSK1/2 QPSK3/4 16QAM1/2 16QAM3/4 64QAM2/3 64QAM3/4

12 10 rcc

8 6 4 2 0 0

1000

2000

3000 R [m]

4000

5000

6000

Fig. 8.8 Reuse co-channel factor as a function of the coverage distance with MCS level as a parameter, in the DL without sub-channelisation

40 BPSK1/2 BPSK3/4

35

QPSK1/2

30

QPSK3/4 16QAM1/2

K

25

16QAM3/4 64QAM2/3 64QAM3/4

20 15 10 5 0 0

1000

2000

3000 R [m]

4000

5000

6000

Fig. 8.9 Reuse pattern as a function of the coverage distance with MCS level as a parameter, in the DL without sub-channelisation

It is therefore important to address techniques for the improvement of system capacity and coverage range. As sub-channelisation in the IEEE 802.16-2004 UL is an optional feature, one explored its impact on the achieved MCS. Differently from the mobile version of WiMAX, IEEE 802.16-2004, based on Orthogonal Frequency division Multiplexing Physical (OFDM-PHY), does not support sub-channelisation in the DL. For the UL, the use of sub-channelisation limits the SS transmissions to 1/16 of the bandwidth assigned to the communication through the BS. The standard defines 16 sub-channels, and 1, 2, 4, 8 or all sets of sub-channels can be assigned to a SS, and each subscriber may use a different MCS in a more permanent way as far he/she is using a different sub-channel.

8 Radio and Network Planning

331

40 BPSK1/2 BPSK3/4 QPSK1/2 QPSK3/4 16QAM1/2 16QAM3/4 64QAM2/3 64QAM3/4

35 30

K

25 20 15 10 5 0 0

1000

2000

3000 R [m]

4000

5000

6000

Fig. 8.10 Reuse pattern as a function of the coverage distance with MCS level as a parameter, in the UL without sub-channelisation

50

BPSK1/2 BPSK3/4 QPSK1/2 QPSK3/4 16QAM1/2 16QAM3/4 64QAM2/3 64QAM3/4

40

M [dB]

30 20 10 0 –10 –20 0

1000

2000

3000 R [m]

4000

5000

6000

Fig. 8.11 Interference-to-noise ratio with sub-channelisation (valid for both links)

Nevertheless, in the DL, as the MCS can be chosen at burst level, there is also the flexibility of using different MCS by different users (even without sub-channelisation), as they use different consecutive bursts within a frame. By using (8.10) and (8.15) one obtains the charts for M(R) and K(R) from Figs. 8.11 and 8.12, respectively. Table 8.5 presents the new values for the vertical asymptotes, which are clearly higher than the ones without sub-channelisation (more than twice the value). In this case, with 16 sub-channels, the variation of the noise power is 10log(1/16) ¼ 12 dB, providing an enhancement of 12 dB in the link budget. By comparing the evolution of K(R) from Fig. 8.12 with the case without sub-channelisation there is no decrease in the reuse pattern; as a consequence, there is no direct increase in system capacity for the lowest values of R (only the achievable coverage distances are increased). For K ¼ 7, the achievable

332

F.J. Velez et al. 40

BPSK1/2 BPSK3/4 QPSK1/2 QPSK3/4 16QAM1/2 16QAM3/4 64QAM2/3 64QAM3/4

35 30

K

25 20 15 10 5 0 0

1000

2000

3000 R [m]

4000

5000

6000

Fig. 8.12 Reuse co-channel pattern as a function of the coverage distance with MCS level as a parameter, in the UL with sub-channelisation

Table 8.5 Values for the vertical asymptote with subchannelisation

Level 1 2 3 4 5 6 7 8

MCS BPSK 1/2 BPSK 3/4 QPSK 1/2 QPSK 3/4 16-QAM 1/2 16-QAM 3/4 64-QAM 2/3 64-QAM 3/4

Rasymptote (m) 14,652.51 12,375.98 11,461.63 9,533.36 7,400.25 5,969.16 4,129.65 3,766.28

MCS is QPSK . However, the achievable coverage distance increases from 2 to 5 km, approximately. It is a worth noting that, in this case, the QPSK ½ MCS may be used up to a coverage distance of 6 km. In order to achieve higher system capacity the use of sectorization is suggested. The use of 120 sectorial BS antennas is adopted, that is, one proposes the use of trisectorial antennas. By inverting both members of (8.9) while using the formula for C/I from [12], one obtains ½ðrcc þ 0:7Þg þ ðrcc  0:22Þg  

1 ¼ 0; ð1 þ M1 Þ  ðC=N Þmin

(8.22)

which is valid both for UL and DL. The minimum reuse factor required for such a tri-sectorial system may be obtained by solving this equation in order to rcc while obtaining the reuse pattern through the use of (8.16). Note that, for the omnidirectional case, the usual assumptions for interference coming from six sources of interference were used in the computations. Although one only have considered

8 Radio and Network Planning

333

25

BPSK1/2 BPSK3/4 QPSK1/2 QPSK3/4 16QAM1/2 16QAM3/4 64QAM2/3 64QAM3/4

20

K

15 10 5 0 0

1000

2000

3000 R [m]

4000

5000

6000

Fig. 8.13 Reuse pattern as a function of the coverage distance with MCS level as a parameter, in the UL with sectorisation but without sub-channelisation

one ring of interference some care would be needed if lower propagation exponents were used [11, 13]. Figure 8.13 presents the chart for the variation of the reuse pattern, K, with the coverage distance. As no sub-channelisation is considered in this case, the vertical asymptotes are the ones from Table 8.4. It is observed that, with sectorization, a clear improvement is obtained in the reuse pattern results in comparison with the ones previously presented, as reuse patterns suffer an important reduction. For K ¼ 7 it is now possible to consider level 6 MCS, that is, 16-QAM 3/4, up to R  1.5 km, overcoming the level 5 MCS without sectorization. While 16-QAM 3/4 can be used for coverage distances up to 1.5 km, 16-QAM ½ may be used up to R  2.2 km. With the QPSK ½ and QPSK 3/4 MCSs it is possible to achieve K ¼ 7 for coverage distances up to approximately 4.1 and 3.3 km, respectively. The use of sub-channelisation increases the coverage distance while the use of sectorization increases the achievable system capacity (through the decrease of the reuse pattern). It is therefore worthwhile to explore the simultaneous use of subchannelisation and sectorization. Figure 8.14 presents the variation of the reuse pattern with the cell coverage distance for this new case, only possible in the UL for fixed WiMAX. Larger coverage distances are possible together with lower reuse distances. With K ¼ 7, it is possible to achieve the 16-QAM 3/4 MCS (level 6) up to R ¼ 4.3 km, while the achievable MCS is 16-QAM ½ MCS (level 5) up to a coverage distance of 6.0 km. In this case, with K ¼ 3, lower order MCSs, for example, BPSK ½, BPSK 3/4, QPSK ½, or QPSK 3/4, are perfectly achieved up to coverage distances larger than 6.0 km. The reduction of the reuse pattern directly corresponds to an increase in system capacity (but not in the cellular coverage). However, it indirectly contributes to an

334

F.J. Velez et al. 25

BPSK1/2 BPSK3/4 QPSK1/2 QPSK3/4 16QAM1/2 16QAM3/4 64QAM2/3 64QAM3/4

K

20

15

10

5

0 0

1000

2000

3000 R [m]

4000

5000

6000

Fig. 8.14 Reuse pattern as a function of the coverage distance with MCS level as a parameter, for the UL, with the use of sectorisation and sub-channelisation

increase in system capacity as higher level MCSs are made available into outer cell coverage rings through the use of adaptive MCSs. One issue that is left for further study is the dependence of these results on the propagation exponent, g. For example, if the propagation exponent decreases the value of the coverage distance asymptote will increase but the asymptotic value for the reuse factor, rcc, will also increase, corresponding to a reduction on the supported system capacity by each MCS.

8.3.4

CNIR and Supported Throughput

To better understand the changes caused by sub-channelisation (16 sub-channels) and sectorization it is worthwhile to plot the CNIR curves as a function of the reuse factor, rcc, with R as a parameter. To produce these curves the power of the carrier is obtained by computing the power received by an SS at a distance R from the BS while the computation of the interference depends on the UL and DL configuration, and on the use of sectorization as well. It can be computed, for a fixed R, by making the same considerations for the reuse as assumed for Equations (8.11), (8.13) and (8.22). Although IEEE 802.16-2004 cannot use sub-channelisation in the DL, one important comparison is between absence and presence of sub-channelisation in the UL. Another important case is the simultaneous use of sub-channelisation and sectorization. From these charts, considering C/N thresholds for each MCS, that is, (C/N)min, it is straightforward to obtain the achievable physical cell throughput, not considering the mixture of services and applications and the corresponding “multiplexing” characteristics. These charts allow for exploring what the achievable CNIR and

8 Radio and Network Planning

335

throughput are for a given reuse pattern, K. As, for hexagonal cells, K ¼ 7 corresponds to a reuse factor rcc ¼ 4.58 this value is considered as a goal. We are aware that, for the cases where sub-channelisation is considered in Fixed WiMAX, the approach we follow for CNIR does not cope with per sub-channel equivalent SINR (or CNIR) computations. These computations can be performed accounting either for exponential effective SINR mapping (EESM), effective code rate map (ECRM), or mean instantaneous capacity (MIC), and may be applied in the future to improve the relevance of the CNIR curves through the use of one of these compression techniques. Figure 8.15 shows CNIR as a function of rcc with R as a parameter for the UL while Fig. 8.16 presents the corresponding variation of the achievable physical throughput with rcc for each MCS.

25

R = 250m R = 500m

20 C/(N + I) [dB]

R = 1000m R = 1500m

15

R = 2000m

10

R = 3000m R = 4500m

5

R = 5000m

0 –5 –10 0

2

4

6

8

10

rcc

Fig. 8.15 CNIR as a function of rcc with R as a parameter, in the UL, no sub-channelisation

12

R = 250m R = 500m

Rb [Mbps]

10

R = 1000m R = 1500m

8

R = 2000m R = 3000m

6

R = 4500m R = 5000m

4 2 0 0

2

4

6

8

10

rcc

Fig. 8.16 Physical throughput as a function of rcc with R as a parameter, in the UL, no subchannelisation

336

F.J. Velez et al.

It is clear that for rcc ¼ 4.58 values of CNIR are always lower than 8.9 dB while for coverage distances larger than 2 km it decreases drastically. As a consequence, for rcc ¼ 4.58, the achievable physical throughput is very low (Rb ¼ 2.82 Mbps maximum) compared with the maximum achievable, that is, 12.27 Mbps, Fig. 8.16. For the DL the results are better, Fig. 8.17, and PHY throughput of 4.23 Mbps is achieved for distances up to 2 km. Figures 8.18 and 8.19 present the results for CNIR and Rb as a function of rcc, in the UL with sub-channelisation. Improvements are only evident for the longest coverage distances, that is, with sub-channelisation the main improvement is on the coverage. If sectorization is applied alone then the values for CNIR at rcc ¼ 4.58 will be higher. However, a truly improvement for coverage distances up to 3 km (not only 2 km anymore) requires both sectorization and sub-channelisation for the UL, as CNIR exceeds 15 dB, as shown in Figs. 8.20 and 8.21. 14

R = 250m

12

R = 500m R = 1000m

Rb [Mbps]

10

R =1500m R = 2000m

8

R = 3000m R = 4500m

6

R = 5000m

4 2 0 0

2

4

6

8

10

rcc

Fig. 8.17 Physical throughput as a function of rcc with R as a parameter, in the DL 25

R = 250m R = 500m

C/(N + I) [dB]

20

R = 1000m

15

R = 1500m

10

R = 3000m

R = 2000m R = 4500m R = 5000m

5 0 –5 –10 0

2

4

6

8

10

rcc

Fig. 8.18 CNIR as a function of rcc, with R as a parameter, in the UL with sub-channelisation

8 Radio and Network Planning

337

12 R = 250m

10

R = 500m

Rb [Mbps]

R = 1000m

8

R = 1500m R = 2000m

6

R = 3000m R = 4500m

4

R = 5000m

2 0 0

2

4

6

8

10

rcc

Fig. 8.19 Supported throughput as a function of rcc, with R as a parameter, in the UL with subchannelisation

30 25

C/(N + I) [dB]

20 15 R = 250m R = 500m R = 1000m

10 5

R = 1500m R = 2000m R = 3000m R = 4500m R = 5000m

0 –5 –10 0

2

4

6

8

10

rcc

Fig. 8.20 CNIR as a function of rcc, with R as a parameter, in the UL with sub-channelisation and sectorization

This leads to a clear confirmation of the need for the simultaneous use of both improvement techniques. Improvements in the analysis can be obtained with the assessment of the supported throughput, Rb, as a function of the distance within a cell when the coverage distance (or cell radius) takes a given value, for example, 2 or 3 km. As, for a fixed rcc, larger coverage distances imply coverage limitations but less interference, it is important to compare the achievable maximum throughput for a given reuse pattern, for example, K ¼ 7, for each value of the distance, d (from the BS to an SS) and each MCS, for different Rs. In the omnidirectional case, while the power of the received carrier is computed for a distance d, the distances assumed for the computation of co-channel

338

F.J. Velez et al. 14

R = 250m R = 500m R = 1000m R = 1500m R = 2000m R = 3000m R = 4500m R = 5000m

12 Rb [Mbps]

10 8 6 4 2 0 0

2

4

6

8

10

rcc

Fig. 8.21 Physical throughput as a function of rcc, with R as a parameter, in the UL with subchannelisation and sectorization

14 R = 3km

Rb [Mbps]

12

R = 2km

10 8 6 4 2 0

0

500

1000

1500 d [m]

2000

2500

3000

Fig. 8.22 Maximum achievable physical throughput as a function of d for R ¼ 2 and 3 km, in the DL in the absence of sub-channelisation and sectorization (K ¼ 7)

interference are maintained, that is, rcc  (R  1) in the UL and rcc  R in the DL. In the tri-sectorial case, the power of the received carrier is also computed for a distance d, whilst considering the following equation for co-channel interference d g C= ¼ g I ðr  R þ 0:7  dÞ þ ðr  R  0:22  d Þg ; cc cc

(8.23)

Figure 8.22 shows the maximum achievable physical throughput as a function of d for K ¼ 7 for the DL, in the absence of sub-channelisation and sectorization, while Fig. 8.23 presents the case of UL, and achieved results are worst. Figure 8.24 shows the same type of results in the presence of sub-channelisation but without sectorization while Fig. 8.25 presents results in the absence of subchannelisation and with sectorization. For a fixed K, with the use of sectorization

8 Radio and Network Planning

339

14 R = 3km

Rb [Mbps]

12

R = 2km

10 8 6 4 2 0

0

500

1000

1500 d [m]

2000

2500

3000

Fig. 8.23 Maximum achievable physical throughput as a function of d for R ¼ 2 and 3 km, in the UL in the absence of sub-channelisation and sectorization (K ¼ 7)

14 R = 3km

Rb [Mbps]

12

R = 2km

10 8 6 4 2 0

0

500

1000

1500 d [m]

2000

2500

3000

Fig. 8.24 Maximum achievable physical throughput as a function of d for R ¼ 2 and 3 km, in the UL, with sub-channelisation but without sectorization (K ¼ 7) 14 R = 3km

12

R = 2km

Rb [Mbps]

10 8 6 4 2 0 0

500

1000

1500 d [m]

2000

2500

3000

Fig. 8.25 Maximum achievable physical throughput as a function of d for R ¼ 2 and 3 km, in the UL, without sub-channelisation but with sectorization (K ¼ 7)

340

F.J. Velez et al.

(tri-sectorial 120 antennas) it is visible a clear increase on the distances where the highest throughputs are achievable. Figure 8.26 presents the case with the simultaneous use of sub-channelisation and sectorization (tri-sectorial 120 antennas). Tables 8.6 through 8.8 present a summary of the values of the maximum distances up which the different MCS may be used (and the respective values of the physical throughput). By using these values for the distance, it is possible to obtain the area of the coverage rings where each MCS is supported in the absence of sectorization, Tables 8.9 and 8.10 (for R ¼ 2 and 3 km, respectively), and with sectorization, Tables 8.11 and 8.12 (in the presence or absence of sub-channelisation, respectively). If one assumes a uniform distribution of users, the area of the coverage rings represents the percentage of use for each MCS. For a fixed K, the use of sub-channelisation improves the coverage slightly but only sectorization clearly improves the results for the achievable physical throughput. Furthermore, the simultaneous use of sub-channelisation and sectorization clearly benefits the possibility of using the highest level MCSs for larger coverage distances (compared with the cases of absence of sub-channelisation and sectorization). 14 R = 3km

12

R = 2km

Rb [Mbps]

10 8 6 4 2 0

0

500

1000

1500 d [m]

2000

2500

3000

Fig. 8.26 Maximum achievable physical throughput as a function of d for R ¼ 2 and 3 km, in the UL, with sub-channelisation and sectorization (K ¼ 7)

Table 8.6 Achievable distance, d, versus supported physical throughput and MCS, for R = 2 and 3 km in the absence of sub-channelisation and sectorization MCS d (m) Rb (Mbps) R ¼ 2 km R ¼ 3 km DL UL DL UL 12.27 64-QAM 910 750 1,185 1,030 11.29 64-QAM 2/3 995 820 1,295 1,130 8.47 16-QAM 1,420 1,190 1,865 1,635 5.64 16-QAM ½ 1,740 1,475 2,295 2,025 4.23 QPSK 2,000 1,900 2,920 2,610 2.82 QPSK ½ – 2,000 3,000 3,000

8 Radio and Network Planning

341

Table 8.7 Achievable distance, d, versus supported physical throughput and MCS, for R = 2 and 3km in the presence of sub-channelisation but no sectorization MCS d (m) Rb (Mbps) R ¼ 2 km R ¼ 3 km DL UL DL UL 12.27 64-QAM – 780 – 1,165 11.29 64-QAM 2/3 – 860 – 1,280 8.47 16-QAM – 1,240 – 1,850 5.64 16-QAM ½ – 1,540 – 2,295 4.23 QPSK – 1,985 – 2,955 2.82 QPSK ½ – 2,000 – 3,000

Table 8.8 Achievable distance, d, versus supported physical throughput and MCS, for R = 2 and 3 km in the presence of sub-channelisation and sectorization MCS d (m) Rb (Mbps) R ¼ 2 km R ¼ 3 km DL UL DL UL 12.27 64-QAM – 1,460 – 2,095 11.29 64-QAM 2/3 – 1,605 – 2,300 8.47 16-QAM – 2,000 – 3,000 5.64 16-QAM ½ – – – – 4.23 QPSK – – – – 2.82 QPSK ½ – – – –

Table 8.9 Percentage of use of each MCS for R = 2 km MCS DL UL Rb (Mbps) d (m) Area (%) d (m) Area (%) 11.29–12.27 64-QAM 995 24.75 820 16.51 5.64–8.47 16-QAM 1,740 50.94 1,475 37.58 2.82–4.23 QPSK 2,000 24.31 2,000 45.61

Table 8.10 Percentage of use of each MCS for R = 3 km d (m) DL Rb (Mbps) 11.29–12.27 5.64–8.47 2.82–4.23

MCS 64-QAM 16-QAM QPSK

d (m) 1,295 2,295 3,000

Area (%) 18.63 39.89 41.48

d (m) 1,130 2,025 3,000

UL&DL with sub-channelization d (m) Area (%) 860 18.48 1,540 40.80 2,000 40.71

UL Area (%) 14.19 31.37 54.44

UL&DL with sub-channelization d (m) Area (%) 1,280 18.20 2,295 40.32 3,000 41.48

Table 8.13 presents the results for the average of maximum throughput, Rb . They are obtained by combining the results for the supported throughput in each coverage ring with the values of the covered area of that crown (which is used as a weight), for both link directions in the presence and absence of sub-channelisation.

342

F.J. Velez et al.

Table 8.11 Percentage of use of each MCS in the absence of sub-channelisation but with sectorization MCS R ¼ 2 km R ¼ 3 km Rb (Mbps) d (m) Area (%) d (m) Area (%) 11.29–12.27 64-QAM 1,295 41.93 1,500 25.00 5.64–8.47 16-QAM 2,000 58.07 2,695 55.70 2.82–4.23 QPSK – – 3,000 19.30

Table 8.12 Percentage of use of each MCS in the presence of sub-channelisation and sectorization MCS R ¼ 2 km R ¼ 3 km Rb (Mbps) d (m) Area (%) d (m) Area (%) 11.29–12.27 64-QAM 1,605 64.40 2,300 58.78 5.64–8.47 16-QAM 2,000 35.60 3,000 41.22 2.82–4.23 QPSK – – – –

Table 8.13 Average of the maximum physical throughput for R = 2 and 3 km R ¼ 2 km R ¼ 3 km Rb (Mbps) No subWith subNo subWith sub channelisation channelisation channelisation channelisation DL UL UL DL UL UL Omnidirectional 7.624 6.473 6.803 6.753 5.886 6.751 Sectorial 9.664 9.664 10.808 7.756 7.756 10.605

In the absence of sectorization, while for R ¼ 2 km the consideration of subchannelisation only leads to an increase in UL throughput of 4.8%, from 6.473 to 6.803 Mbps, for R ¼ 3 km, the increase achieves 15.1%, from 5.886 to 6.751 Mbps. While for R ¼ 3 km the use of sub-channelisation leads to an almost perfect balance between the DL and UL, for R ¼ 2 km this improvement does not occur. This can also be verified by comparing the right hand columns of Tables 8.9 and 8.10. With sectorization, while for R ¼ 2 km the consideration of sub-channelisation leads to an increase in throughput in the UL of 11.84%, from 9.664 to 10.808 Mbps, for R ¼ 3 km, the respective increase is 36.73%, from 7.756 to 10.605 Mbps. In this case, with the use of sub-channelisation, UL throughput surpasses DL throughput. In the omnidirectional case, with the use of sub-channelisation, although the noise power decreases the improvement only allows for a perfect balance of the PHY throughputs between the UL and DL for R ¼ 3 km (but not for R ¼ 2 km). The use of sectorization reduces the interference, and an improvement on the supported physical throughput of 59% is obtained for R ¼ 2 km (from 6.803 to 10.808 Mbps). However, for R ¼ 3 km, as the impact of noise is higher, without the use of sub-channelisation, the improvement only reaches 32% (from 5.886 to 7.756 Mbps). For R ¼ 3 km, only with the use of sub-channelisation and sectorization an improvement of 57% is achieved (from 6.751 to 10.605 Mbps).

8 Radio and Network Planning

8.3.5

343

Summary and Conclusions

From the analysis it is clear that both noise and interference present a strong limitation to the performance of fixed WiMAX, mainly for higher level MCSs. As a consequence, with a reuse pattern K ¼ 7, cell throughputs near the maximum are only achieved in the UL if sub-channelisation is used together with sectorization. For lower coverage distances, the use of sectorization alone in the UL allows for a substantial gain in the physical throughput. However, for larger coverage distances, in the absence of sub-channelisation the achieved gain is not comparable with the case where sub-channelisation is used. These results motivate that future research directions need to be explored to analyse the inclusion of sub-channelisation into the DL, as the IEEE 802.16e standard already supports. Although this will give extra flexibility on resource assignment the extra complexity in the dimensioning process will justify the need for new methodologies, for example, considering scheduling. Another conclusion that can be extracted from the work is the need to improve the results for the maximum physical throughput in the outer crowns of the cells as this is the zone that suffers the highest interference. The use of relays within fractional reuse schemes, as shown in Fig. 8.27, is being pointed out as a solution for this challenge, and it needs to be investigated. The presence of relay stations (either fixed or mobile), with limited coverage, will introduce new challenges into the design process, as interference can be mitigated in different ways whilst increasing coverage.

Fig. 8.27 Fractional frequency reuse in a relaybased system where the number of fractional relay stations is 3 (extracted from [14, 15])

Base station Relay station

344

8.4

F.J. Velez et al.

Fixed WiMAX Planning Platforms

8.4.1

Framework and Scenario

Models for the achievable physical throughput, cellular coverage, and frequency reuse are very useful for the automation of cellular planning procedures. Cellular planning is highly dependent on the propagation environment and a careful choice of the placement, height and tilt of the BS transmitter antennas is needed in order to ensure a high percentage of LoS within the cells. As a consequence, the use of Geographic Information System (GIS) is needed to account for the terrain profile. In the context of the MobileMAN project [16], a cellular fixed WiMAX PtM network [1] was deployed, covering the whole district of Covilha˜, and in particular the city area. While the overall cellular structure is mainly dedicated to emergency and security public services, urban micro-cells will support e-learning and e-health services, among others. Although the district of Covilha˜ area is 550 km2, the territorial framework of the frequency band license assigned by ANACOM, the Portuguese regulator, is broader, and includes the whole area under study within the MobileMAN project in Beira Interior, Portugal, where a cellular planning exercise was also performed. A GIS based WiMAX planning tool was conceived that includes aspects of coverage, frequency reuse, and the impact of classes of services and applications. The tool enables radio and network planning of WiMAX outdoor networks. It relates the geographical data of a given location to the number of BSs needed to cover that location. This WiMAX platform uses ArcGIS as a working environment [17]. ArcGIS allowed the development of the toolbar that works as a base for network planning. The radio characteristics of the system are studied, including the link budget, the radio capacity, and the definition of the radio propagation models. Although our tool does consider adaptive MCSs it did not include the functionality of distinguishing among different MCSs in the graphical presentation of the results, that is, it does not display different maps for each MCS. It is therefore worthwhile to explore the functionalities of commercially available tools. The choice was WinpropTM from AWE, which considers the information coming from the digital terrain profile (as our tool does) but displays the several different MCSs on different maps.

8.4.2

Wireless Planning Tool

8.4.2.1

Functionalities and Potentialities

The planning tool is made available as a toolbar in ArcGIS that allows for the user to choose several parameters, from the number of base stations to the number of

8 Radio and Network Planning

345

users, allowing for network operation in different environments, providing important similarities with real-life situations. The inputs of the tool are the file with the map, including the elevation of the terrain, in digital format, the total coverage area, the most probable position of SSs, as well as the type of applications. The options included into the ArcGIS toolbar are accessed via buttons, Fig. 8.28, and include the following WiMAX planning functionalities: l

l

Definition of urban zones – With this button urban zones are defined for BSs not operating at the maximum transmitter power since, for these sites, as the user density is much higher than in rural areas, the limitation is the system capacity instead of the radio coverage. Figure 8.29 presents an example of the urban zone of Covilha˜ over the Digital Map Terrain (DMT). Equipment – There are two buttons that deal with the WiMAX equipment to be used. The first enables to choose the type of equipment while the second enables the visualization of the equipment data. The equipment list may include items either from a manufacturer available on the market, or customized by the user definitions. In the former case, it is possible to define the bandwidth, frequency, and type of antenna. In the latter, the planner may define all the parameters, for example, bandwidth, transmission power, sensitivity, frequency, and type of antenna, Fig. 8.30.

Fig. 8.28 Toolbar on the graphic environment of ArcGIS

Fig. 8.29 Urban zones in the digital map terrain

346

F.J. Velez et al.

Fig. 8.30 Equipment characteristics: example of a possible choice for the parameters, different from the manufacturer ones

Fig. 8.31 BS positioning and LoS and NLoS areas

l

l

Sites with LoS and NLoS – This option allows for verifying the existence of LoS, and helps to choose the propagation model to be used on each zone of the map, compute the received power and analyze the type of application. This verification is performed by one of the GIS tools, and allows for an optimum verification of the LoS regions originated from each BS, as depicted in Fig. 8.31, example for the district of Covilha˜. The actual ArcGIS function used in the tool is Viewshed, a 3D Analyst extension of ArcView. Received power and coverage – This functionality creates three different layers. One has the coloured map for each value computed for received power (at each point of the map), depicted in Fig. 8.32. The other shows the values for CNIR. By considering received power and CNIR values, it is possible to check the areas with appropriate coverage and select the spatial option for the capacity. The last layer is a scheme where the different areas covered by each BS can be visualized, as well as the interference areas and the areas with no coverage.

8 Radio and Network Planning

347

Fig. 8.32 Map with different kind of users (district of Covilha˜)

Fig. 8.33 Definition of the distribution of users by classes, their real-time requirements and the density

l

Users – This button shows a menu to define the percentage of users in each class (e.g., percentages of users with real time services) and the density of users per km2, as shown in Fig. 8.33. Figure 8.32 shows that these users are then randomly distributed on the map, whilst distinguishing the different classes by different symbols.

The service classes (labelled as one, two and three), distinguish real-time applications at 64, 384, and 2,000 kbps, respectively. A given percentage of users has access to time-based applications (the ones where the time is an intrinsic component of the application), for example, voice or video [18], while others are using non time-based ones. Figure 8.34 depicts some of the results, namely the distribution of users by classes, percentage of area covered and subject to interference, MCSs for each BS, number of users at each cell, throughput and percentage of served users.

348

F.J. Velez et al.

Fig. 8.34 Results arising from the ArcGIS planning tool

Fig. 8.35 LoS coverage for the district of Covilha˜

8.4.2.2

Covilha˜ Rural and Urban Areas

LoS discovery should be applied for a better cellular planning. GIS functionalities were incorporated into the tool for the choice of the best placement of BSs, including their height. By considering the use of the modified Friis model, an initial application was made for the district of Covilha˜, an area of 550 km2. Because this zone is very hilly, cells with coverage distances around 3 km are used, differently from the whole region of Beira Interior, where larger cells were considered. By considering 18 BSs and by using digital terrain models and ArcGIS 9.0, 3D Analyst extension, one obtains LoS coverage in 70% of the area, Fig. 8.35. In this first case, omnidirectional antennas are mounted on 15 sites (although two of them only cover a 180 sector), while the remaining three sites have two 180 sectorial antennas. It was verified that there is an average of 83% LoS coverage in villages,

8 Radio and Network Planning

349

Fig. 8.36 Use of omnidirectional antennas in the district of Covilha˜

Fig. 8.37 Use of sectorial antennas in the district of Covilha˜

towns and cities. This choice guarantees propagation exponents of g ¼ 2 in rural areas and g ¼ 3 in dense urban areas. Furthermore, main roads on the access to the mountain are covered with LoS. Another exercise using the ArcGIS platform, in a broader geographical scope, leads to the results from Figs. 8.36 and 8.37, where micro-cells are overlaid with the macro-cellular structure. The height of BS antennas is 30 m, and slightly higher antenna gains are considered compared to the experimental setup. Besides, the transmitter power in urban areas is 18 dB lower than in rural areas. The SUI-C model was considered in this second case. A comparison between the use of omnidirectional and tri-sectorial cells is performed. In terms of interference mitigation, the advantage of using sectorial antennas is clear. From Table 8.14 one can observe that, with sectorial antennas, the area of interference is reduced from 42%, in the omnidirectional case, down to 9.3%, while the covered area (without interference) increases from 52.3% to 85.0%.

350

F.J. Velez et al.

Table 8.14 Coverage and interference areas for the district of Covilha˜ Type of Antenna Coverage area (%) Without interference With interference (%) Omnidirectional 52.3 42.0 Sectorial 85.0 9.3

Non-covered Area (%) 5.7 5.7

Fig. 8.38 Cellular coverage using sectorial antennas in the whole region of Beira interior

8.4.2.3

Region of Beira Interior

For the WiMAX cellular coverage of the whole region of Beira Interior, a zone with an area of approximately 5,760 km2 was considered while exploring many combinations for the placement of BSs and types of antennas. One specific exercise compares the use of omnidirectional antennas with the use of sectorial ones over the whole region (Fig. 8.38). The advantages of using the latter in terms of interference mitigation are clear because the area of interference is reduced by 36.1%, from 36.4% to 0.3% while the covered area increases from 50.8% to 86.9%. Finally, it is worthwhile to note that, as the SUI model is pessimistic in terms of propagation, the results for coverage are appropriate from an engineering perspective, as they represent a worst-case situation. However, results for interference would be worse if a more optimistic model was used for propagation at 3.5 GHz. A strong need of using sectorial antennas is verified. Sectorial cells guarantee an adequate coverage and interference mitigation for several terrain types and environments, including hilly terrains. In the particular case of our experiments, optimal planning for the integration of WiMAX and Wi-Fi technologies enables real technical conditions to make the interoperability of these HSF networks available to the students of the Health Science Faculty of University of Beira Interior, where the demonstrator is placed. It provides excellent theoretical, laboratorial, simulation and practical lessons

8 Radio and Network Planning

351

through multimedia and IP communications, for example, videoconference, voice over IP, and communication of high resolution video/image with the support of terminal mobility, a must for the practice and teaching of medicine. As a final note it is important to mention that these technologies are very attractive in economic terms, when compared to legacy network technologies.

8.4.3

Results with Adaptive Modulation and Coding Schemes

8.4.3.1

Scenario

In the cellular planning exercises with the WinpropTM tool, the geographical scope and the location of BSs were the same as previously considered in Fig. 8.35. However, in this case, 18 tri-sectorial BSs are considered in the district of Covilha˜, which corresponds to a total 54 sectors, Fig. 8.39. WinpropTM enables to present different maps distinguishing each MCS. It also facilitates to explore other propagation models [8], for example, the dominant path loss one, and actual antenna pattern characteristics (from a file with the manufacturer characteristics). In this part of the work, one considers the parameters from the WiMAX BreezeMAX Alvarion equipment [9], and the frequency bands assigned by ANACOM. The BreezeMAX 3,000 equipment is the Alvarion WiMAX platform for the 3–4 GHz licensed bands. Table 8.15 shows the parameters used in the planning.

Fig. 8.39 Topography of the district of Covilha˜ covered with 18 base stations

352

F.J. Velez et al.

Table 8.15 Parameters for the planning

Air interface Number of cell Number of sector per cell Operation frequency Duplexing mode Multiple access technique Bandwidth DL/UL separation Antenna pattern BSs height SS height Maximum BS output power BS antenna gain SS antenna gain BS Noise Figure SS Noise Figure Propagation model

IEEE 802.16-2004 18 3 3.5 GHz FDD TDMA 3.5 MHz 100 MHz WiMAX 3500 MHz 120 Several, according to the topography 1.5 m 43 dBm (19.95 W) 15.3 dBi 1 dBi 4 dB 7 dB Dominant path loss

Fig. 8.40 Zones with LoS to BS15

8.4.3.2

Propagation in the District of Covilha˜

As an example, Fig. 8.40 presents the LoS regions seen from BS 15 in red while the NLoS regions are presented in green. BS15 is located at the following coordinates: X ¼ 247,000 m, Y ¼ 369,700 m (whose reference is a point on the sea near Sagres Point in the south of Portugal), the one assumed by the Portuguese Army Surveying Institute. The altitude for BS 15 is 1758 m, near the highest point in Portugal continental, the Torre (1993 m), and the antenna tower height is 20 m. One verifies that there is

8 Radio and Network Planning

353

Fig. 8.41 Prediction for the received power for BS15

–45

Power [dBm]

–50 –55 –60 –65 –70 –75 –80

0

1000

2000

3000

4000

5000

6000

7000

Distance [m]

Fig. 8.42 Curve for the received power versus the distance (extracted from the WinpropTM tool)

LoS to the highest regions of the district of Covilha˜, and there only is NLoS to the down region, near the city of Covilha˜. Figure 8.41 shows a prediction of the received power at each of the three sectors of BS15. Figure 8.42 is also extracted from the WinpropTM tool in an actual case and helps to analyse the variation of the received power with the distance. The arrow represents the orientation of the sector.

354

F.J. Velez et al.

Table 8.16 Correspondence between the MCSs, CNIR and transmission modes

8.4.3.3

ID

MCS

CNIRmin (dB)

1 2 3 4 5 6 7 8

BPSK 1/2 BPSK 3/4 QPSK 1/2 QPSK 3/4 16-QAM ½ 16-QAM 3/4 64-QAM 2/3 64-QAM 3/4

3.3 5.5 6.5 8.9 12.2 15.0 19.8 21.0

Physical throughput (Mbps) 1.41 2.12 2.82 4.23 5.64 8.47 11.29 12.27

Influence of the Adaptive Modulation and Coding Scheme

In the dimensioning process, it is important to analyse the spatial dependence of the available MCS on the signal strength. The WiMAX equipment uses eight different MCS with different CNIR thresholds (Table 8.16) corresponding to the values for the sensitivity presented in Table 8.2. Planning and network performance results are based on the predictions for the received power (or attenuation) that occurs for each BS sector. According to AWE WinpropTM terminology, WiMAX network design is essentially based on the detailed analysis of the spatial predictions for each MCS, for example, CNIR, received power at the SS and BS, number of channels. Based on these results, the tool determines the physical throughput in the different geographical zones.

8.4.3.4

Prediction of the Maximum Received Power

The tool defines a so-called maximum received power. Figure 8.43 presents its prediction for the DL for the whole district of Covilha˜. In the results shown in Fig. 8.43 all the BSs are simultaneously considered in the prediction of the maximum received power, leading to the results for the variation of the maximum received power with the distance from Fig. 8.44. One verifies that the decrease of the maximum received power is much smoother than the variation of the received power itself (represented in Fig. 8.42). Another approach may be to analyse the coverage through the consideration of a threshold for the received power. Figure 8.45 presents this analysis for a received power threshold of 80 dBm. A pixel is red if the power is above the threshold and is green if the power is below the threshold. One verifies that with the 18 sectorial BSs a high percentage of coverage is obtained.

8.4.3.5

Prediction of the Maximum Physical Throughput in the DL

Figure 8.46 presents the Prediction of the PHY Throughput in the DL in the whole area covered by the WiMAX network. These results can be analyzed to decide

8 Radio and Network Planning

355

Fig. 8.43 Prediction for the maximum received power in the DL

Max. Received [dBm]

–40 –45 –50 –55 –60 –65 –70 0

500

1000

1500

2000

2500

3000

3500

Distance [m]

Fig. 8.44 Curve for the prediction for the maximum received power as a function of the distance (extracted from the WinpropTM tool)

where to expect the maximum and the minimum bit rate. For the SSs closer to the BS, an higher order modulation, that is, 64-QAM, is used, whose corresponding bit rate is 12.27 Mbps. For the users located further away, at a medium distance (still not far away from the BS), the 16-QAM modulation is used, and the bit rate is 8.47 Mbps. For the most distant users, near the cell boundary, QPSK or BPSK modulations are used (supporting data rates of 4.23 and 2.12 Mbps, respectively). This means that, as expected, the farthest the SS is from the BS the lowest is the physical

356

F.J. Velez et al.

Fig. 8.45 Coverage by considering a threshold for the maximum received power

Fig. 8.46 Prediction of the maximum reachable PHY throughput in the DL

throughput. To better interpret these results, it is important to analyze the charts from Figs. 8.47, 8.48, and 8.49, which represent the maximum physical throughput versus the distance, the histogram of the maximum physical throughput, and the cumulative probability for the physical throughput, respectively.

8 Radio and Network Planning

357

Max Bit Rate [KBit/s]

12000 10000 8000 6000 4000 2000 0

1000

2000

3000 4000 Distance [m]

5000

6000

Fig. 8.47 Maximum physical throughput versus the distance (extracted from WinpropTM)

0.35 0.30 Probability

0.25 0.20 0.15 0.10 0.05 0.00

0

5000

10000 15000 Max Bit Rate [kBit/s]

Fig. 8.48 Histogram for the maximum bit physical throughput, Rb (extracted from WinpropTM)

By assuming ideal propagation conditions, for distances up to 1.5 km, WiMAX links support the highest order modulation, which corresponds to bit rates that vary between 11.29 and 12.27 Mbps. Then, for distances up to 2.7 km, the order of modulation are the medium ones (bit rates from 5.64 to 8.47 Mbps) while for distances of approximately 6 km one only achieves the lowest order modulations, with bit rates varying from 1.41 to 4.23 Mbps. From Fig. 8.48 one concludes that the MCS mostly used is QPSK (4.23 Mbps), in approximately 35% of the cases, followed by the medium order modulation, 16-QAM (8.47 Mbps), in more than 20% of the cases. The highest order modulation is used with a frequency of 15%. Figure 8.49 represents the cumulative effect of the results from Fig. 8.48.

358

F.J. Velez et al.

Cummulated Probability

1.0 0.8 0.6 0.4 0.2 0.0

0

5000

10000 15000 Max Bit Rate [kBit/s]

Fig. 8.49 Cumulative histogram for the different supported maximum physical throughput (extracted from WinpropTM)

8.4.3.6

Prediction of “Best Server” Cells

In the representation of the “best server” each pixel (representing the SSs) is assigned to a BS sector, more precisely to the carrier assigned to the respective sector, by choosing different colours. As sectorial antennas are used, an aperture angle is defined to each sector antenna. If the pixel is within the range of two different apertures, the assignment is performed to the sector/BS with the lowest associated distance. Figures 8.50 and 8.51 present the maps with the predictions for the “best server” for each MCS. When the MCS level increases, the area of the “best server” cell/sector decreases as it is only possible to use the highest order MCSs in the regions closest to the BSs. The tool is also able to represent the number of received carriers and of received channels (number of carrier multiplied by the number of time slots). A carrier is received if the received power in a SS is above a pre-defined threshold. The number of received carriers is important to address handover issues, as it is easy to identify the number of different solutions to perform handover. However, as the fixed WiMAX does not support handover the results are not presented. 8.4.3.7

Carrier-to-Noise-plus-Interference Ratio in the DL

Maps with the prediction of CNIR (or SNIR, signal-to-noise-plus-interference ratio) in the DL are also individually produced for each MCS, Fig. 8.52. If the received CNIR is below the threshold established to each MCS the pixel is transparently filled; otherwise, it is filled with the colour represented on the scale. The WinpropTM tool does not consider the UL. Although the results for the UL are generally less favourable, the use of sub-channelisation may allow for overcoming these coverage limitations, as previously discussed.

8 Radio and Network Planning

359

Fig. 8.50 Maps with the “best server” sectors for (a) BPSK ½ and (b) BPSK

8.4.4

Summary and Conclusions

Two different approaches were considered for cellular planning in the district of Covilha˜. On the one hand, one considered a GIS based WiMAX planning tool conceived by considering coverage issues, frequency reuse, the impact of the

360

F.J. Velez et al.

Fig. 8.51 Maps with the “best server” sectors for the remaining MCS

different classes of service. On the other, as adaptive MCS are considered WinpropTM was used. The GIS functionalities allow for appropriately adjusting azimuth and tilt of antennas. This cellular planning exercises confirm the results theoretical analysis, where different crowns are achieved for the coverage with each MCS (corresponding to a given range of values for SNIR), and also for the maximum PHY throughput and for the “best server” cells. The benefit of using sectorization was also demonstrated. Cost/revenue optimisation will allow for finding the optima for the planning. There are fixed costs (e.g., spectrum licenses), plus costs proportional to the number of cells, and the cost proportional to the number of “transceivers”. Typically, total costs depend on the size of the cells and on the reuse pattern. Revenues depend on the supported throughput (which also depends on the size of the cells and on the

8 Radio and Network Planning

361

Fig. 8.52 Maps with the SNIR prediction in the DL

reuse pattern). They depend on prices and will be very sensitive to the number of supported users. Regarding the profit, the absolute profit is usually proportional to the supported “traffic” (e.g., per km2). The profit in percentage is, however, proportional to the

362

F.J. Velez et al.

“spectral efficiency”, and companies usually want to maximize the profit in percentage [19]. Further work is needed to seek for optimum coverage distances in different reuse configurations accounting for cost-benefit optimisation.

8.5

Conclusions

In this Chapter, one started by considering a PtM demonstrator for fixed WiMAX that allowed for extracting measurements for cellular coverage in the suburban area of Covilha˜, Portugal. For this zone, and for some ranges for the distances, by using a curve fitting approach one concluded that the modified Friis model can be used with a propagation exponent g ¼ 3. From the analysis of the results, one can also conclude that the Stanford University Interim model can be considered (mainly SUI-C but also SUI-B), as the mean square error is kept under reasonable values. Then, an analytical approach was used to determine the trade-offs between the coverage distance and interference minimisation whilst increasing system capacity. From the analysis, it is clear that both noise and interference present strong limitations to the performance of fixed WiMAX, mainly for higher MCS levels. With a reuse pattern K ¼ 7, cell throughputs near the maximum are only achieved, in the UL, if sub-channelisation is used together with sectorization. With the use of sub-channelisation alone, although the noise power decreases, the improvement is not so clear. For the shortest coverage distances, the use of sectorization alone in the UL allows for obtaining a substantial gain in the physical throughput. However, for larger coverage distances, in the absence of sub-channelisation, the achievable gain is not comparable with the case where sub-channelisation is used. In general terms, the use of sectorization in fixed WiMAX enables to reduce the reuse pattern while considering sub-channelisation allows for improvement on the coverage. The reduction of the reuse pattern directly corresponds to an increase in the system capacity but the improvement in the coverage range (through subchannelisation) can also allow for an improvement in UL system capacity, as adaptive MCS are used. The need to improve the results for the maximum achievable throughput in the outer coverage rings of the cells, as this is the zone that suffers the highest interference, motivates future research directions, for example, by using relays within fractional reuse schemes. The presence of relay stations (either fixed or mobile), with limited coverage, will introduce new challenges into the design process, as interference can be mitigated in different ways whilst increasing coverage (e.g., by decreasing the transmitter power or by using advanced scheduling techniques). One interesting aspect which is left for further study is on the dependence of the results on the propagation exponent, g. For example, if the propagation exponent decreases, for example, to g ¼ 2.3, the value of the asymptote for the coverage area

8 Radio and Network Planning

363

will increase and the asymptotic value for the reuse factor, rcc, will increase, corresponding to a reduction on system capacity for each MCS. Two different approaches were considered for graphical cellular planning, and the district of Covilha˜ was considered as a case study. On the one hand, one considered a GIS based WiMAX planning tool conceived by considering coverage issues, frequency reuse, and the impact of the different classes of service. On the other, WinpropTM was used as it distinguishes among different MCS in the graphical presentation of the results. Both tools consider the information coming from the digital terrain profile. The GIS functionalities allow for appropriately adjusting azimuth and tilt of antennas. This cellular planning exercises confirm the results of theoretical analysis, where different crowns are achieved for the coverage with each MCS (corresponding to a given range of values for SNIR), for the maximum physical throughput, and for the “best server” cells. The frequency radio resources should be considered as the most valuable resource during the planning of wireless broadband access networks. As a rule, spectral efficiency needs to be optimized by using several advanced techniques, corresponding to an optimization from the cost-benefit point of view. Acknowledgement This work was partially funded by MobileMAN (Mobile IP for Broadband Wireless Metropolitan Area Network), an internal project from Instituto de Telecomunicac¸o˜es/ Laborato´rio Associado, by CROSSNET (Portuguese Foundation for Science and Technology POSC project with FEDER funding), by “Projecto de Re-equipamento Cientı´fico” REEQ/1201/ EEI/ 2005 (a Portuguese Foundation for Science and Technology project), and by the Marie Curie Intra-European Fellowship OPTIMOBILE (Cross-layer Optimization for the Coexistence of Mobile and Wireless Networks Beyond 3G, FP7-PEOPLE-2007-2-1-IEF). The authors acknowledge the fruitful contributions on ArcGIS tools from Engº Jose´ Roma˜o, Engº Jose´ Riscado and Prof. Victor Cavaleiro from STIG-UBI, and to the final year project students Hugo Carneiro, Jorge Oliveira, Dany Santos and Rui Marcos.

References 1. IEEE, Draft IEEE Standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed Broadband Wireless Access Systems, IEEE 802.16-REVd/D5, The Institute of Electrical and Electronics Engineers, New York, USA, May 2004 2. IEEE, Draft IEEE Standard for Local and Metropolitan Area Networks – Part 16: Air Interface for Fixed Broadband Wireless Access Systems – Amendment for Physical and Medium Access Control Layers for Combined Fixed and Mobile Operation in Licensed Bands, IEEE 802.16e/D9, The Institute of Electrical and Electronics Engineers, New York, USA, June 2005 3. http://www.awe-communications.com 4. V. Erceg et al., An empirically based path loss model for wireless channels in suburban environments. IEEE J. Select. Areas Commun. 17(7), 1205–1211 (July 1999) 5. IEEE 802.16 Working Group, Channels models for fixed wireless applications, Document 802.16.3c-01/29r4 (July 2001) 6. K. Hari, Interim Channel Models for G2 MMDS Fixed Wireless Applications, in IEEE 802 plenary meeting, Tampa, USA, Sept 2000. www.ieee802.org/16/tg3/contrib/802163c00_49r2.pdf (March 2010)

364

F.J. Velez et al.

7. H.R. Anderson, Fixed Broadband Wireless Systems Design (Wiley, Chichester, West Sussex, UK, 2003) 8. R. Wahl, O. St€abler, G. Wo¨lfle, Propagation model and network simulator for stationary and nomadic WiMAX networks, in Proceedings of IEEE VTC 2007 Fall – IEEE 66th Vehicular Technology Conference, Baltimore, MD, USA, Sept 2007 9. http://www.alvarion.com 10. T.S. Rappaport, Wireless Communications: Principles and Practice (Prentice Hall, Upper Saddle River, NJ, 2002) 11. F.J. Velez, L.M. Correia, J.M. Bra´zio, Frequency reuse and system capacity in mobile broadband systems: comparison between the 40 and 60 GHz bands. Wireless Pers. Commun. 19(1), 1–24 (Aug 2001) 12. F.J. Velez, V. Carvalho, D. Santos, R.P. Marcos, R. Costa, P. Sebastia˜o, A. Rodrigues, Planning of an IEEE 802.16e network for emergency and safety services, in Proceedings of 3G 2005 – 6th IEE International Conference on 3G Mobile Communication Technologies, London, UK, Oct 2005 13. F.J. Velez, Aspects of cellular planning in Mobile Broadband Systems, PhD thesis, Instituto Superior Te´cnico, Universidade Te´cnica de Lisboa, Lisbon, Portugal (Dec 2000) 14. K.A. Rizvi, Young sun, D. Basgeet, Z. Fan, P. Strauch, Fractional frequency reuse for IEEE 802.16j relaying mode. IEEE C80216j-06_223, IEEE (Nov 2006) 15. http://www.wimaxforum.org/documents/downloads 16. http://www.e-projects.ubi.pt/mobileman 17. H. Carneiro, J. Oliveira, A. Rodrigues, P. Sebastia˜o, Software planning tool for WiMAX networks, in Proceedings of Conftele’ 2007 – 7th Conference on Telecommunications, Peniche, Portugal (May 2007) 18. F.J. Velez, L.M. Correia, Mobile broadband services: classification, characterisation and deployment scenarios. IEEE Commun. Mag. 40(4), 142–150 (Apr 2002) 19. F.J. Velez, L.M. Correia, Optimisation of mobile broadband multi-service systems based in economic aspects. Wireless Netw. 9(5), 525–533 (Sept 2003)

Chapter 9

System Capacity Fernando J. Velez, M. Kashif Nazir, A. Hamid Aghvami, Oliver Holland, and Daniel Robalo

Abstract In Fixed WiMAX, the contribution from each transmission mode can be incorporated into an implicit formulation to obtain the supported throughput as a function of the carrier-to-interference ratio. This is done by weighting the physical throughput in each concentric coverage ring by the size of the ring. In this paper, multi-hop cells are formed by a central coverage zone and three outer coverage zones served by cheaper low-complexity relays. It is assumed that line of sight propagation to the bases station is achieved in a high percentage of the cell, reducing the impact of selective fading, through allowing dimensioning to be done by GIS cellular planning tools. By using tri-sectorised equipment there is a need for three times more bandwidth, while hardware costs are higher. In our proposal for relays, the FDD mode is considered and the frames need to guarantee resources for BS-to-MS communications but also for BS-to-RS and RS-to-MS communications. These requirements leads to a 1/5 asymmetry factor between the UL and DL in the omnidirectional BS case and to a 3/7 asymmetry factor in the case of tri-sectored BSs. Although the reuse distance is augmented by a factor pffi 3, we show that with the use of relays in FDD mode only the consideration of tri-sectored BSs with reuse pattern K = 3 (at the cost of extra channels, corresponding to 9 channels) enables to obtain values for the throughput comparable to cases without the use of relays. The presence of sub-channelisation only improves the results for the highest values of R. The consideration of tri-sectored BS antennas with K = 1 (whilst keeping the number of required channels – equal to 3) did not enable to obtain values of the throughput comparable to the ones without using relays, although frame format is more favourable. Relays can be cheaper than BS with full functionalities. As the use of relays may lead to lower costs it is worthwhile to analyse the impact of using them on costs and revenues.

F.J. Velez (*) Instituto de Telecomunicac¸o˜es-DEM, Universidade da Beira Interior, Calc¸ada Fonte do Lameiro, 6201-001, Covilha˜, Portugal e-mail: [email protected]

R. Prasad and F.J. Velez, WiMAX Networks, DOI 10.1007/978-90-481-8752-2_9, # Springer ScienceþBusiness Media B.V. 2010

365

366

9.1

F.J. Velez et al.

Introduction

In the context of Worldwide Interoperability for Microwave Access (WiMAX) planning, research on the variation of the carrier-to-noise-plus-interference ratio (CNIR), against different system parameters, is of fundamental importance. As there are challenges in both the uplink (UL) and downlink (DL) in WiMAX, techniques such as sub-channelisation may be applied to reduce the impact of noise on link performance. However, only Mobile WiMAX allows for sub-channelisation in both the UL and DL; fixed WiMAX only allows for it in the UL this absence of sub-channelisation in the DL for fixed WiMAX may be a cause of performance degradation (mainly owing to the extra noise caused by the larger spectrum bandwidth). For cellular planning purposes, the UL and DL CNIRs from/at the wireless Subscriber Station (SS) are very significant parameters. From a detailed analysis of CNIR variation for different coverage and reuse distances, an evaluation of the achievable reuse patterns can be performed for different modulation and coding schemes (MCSs). In order to more effectively use radio frequency spectrum, it is important to choose a frequency reuse scheme that leads to coverage guarantee and improved system capacity whilst minimising interference. Broadband wireless access enabling the operation of multi-hop relay stations (RSs) aims not only to enhance the coverage but also the system capacity, as interference is mitigated owing to the lowest transmitter power associated to the small range of the RSs. Compared with base stations (BSs), RSs does not need a wire-line backhaul and has much lower hardware complexity; hence, using RSs can significantly reduce the deployment cost of the system. The main objective is to achieve the highest values for the carrier-to-interference ratio, C/I, and, in return, the maximum supported throughput, by using relays for a given frequency reuse pattern, for example, K ¼ 3. In this work, a comparison of the different values the throughput is performed between the RSs, BSs and SSs in topologies with relays. By weighting the physical throughput in each concentric cell coverage ring by the size of the ring, the contribution from each transmission mode (or MCS) is included in an implicit function formulation to obtain the average supported throughput. For consecutive MCSs, the step distances are determined by the correspondence between minimum values at the CNIR curves (for a given MCS) and the supported physical throughput by an inversion procedure (via the consideration of each MCS stepwise threshold). The remaining of this Chapter is organized as follows. Sections 9.2 and 9.3 presents the formulations and assumptions for the CNIR analysis in the DL and UL, for configurations without and with relays for fixed WiMAX configurations. The supported physical throughput is analyzed in Section 9.4, envisaging the cases without relays and with relays (DL and UL) Section 9.5 addresses the special case of unitary reuse pattern, where the use relays may present a slight advantage. Finally, Section 9.6 presents the conclusions.

9 System Capacity

9.2

367

CNIR Versus Physical Throughput Without Relays

In Fixed WiMAX, the supported physical user throughput is a function of the supported MCS, which in turn depends on the achievable CNIR compared with the minimum CNIR, CNIRmin, for each MCS (see Tables 8.2 and 8.16). It is therefore important to analyse the evolution of the CNIR against choices of several system parameters as well as the chosen co-channel reuse factor. If Frequency Division Duplexing (FDD) is used, analytical modelling of coverage and frequency reuse problems can only be carried out in Fixed WiMAX. Our chosen approach simultaneously accounts for carrier-to-noise and carrierto-interference constraints [1, 2]. Figure 9.1 presents the distance associated with coverage and interference for a 2D geometry with six interferers, when the mobile user is at a distance d from its serving base station (BS). The worst-case scenario in the DL occurs when the BS of the serving cell transmits to the most distant possible location of subscriber station (SS) it is serving, using a channel (or sub-channel) on which the SS is also receiving interference from the BSs of the six co-channel hexagonal neighbouring co-cells. If d is replaced by R the cell coverage distance or radius (0  d  R), this worst case is as depicted in Fig. 9.1. Note that if D is the reuse distance, there are tiers of interference at distances D, 2D, etc. However, if a high value for the propagation decay exponent is set, it is a valid approximation to only consider the first tier of interference, as shown in Fig. 9.1. In the UL, the worst-case scenario occurs when the SS is transmitting to the BS from the cell edge, while interfering mobiles are on the boundary between interfering cells’ edges and the serving cell of the SS [3]. When sectorization is considered, the number of interfering cells is decreased and system capacity increases. Such coverage and reuse geometries are commonly found in rural and suburban environments. In urban areas, owing to the obstruction of buildings and other urban obstacles, perfect circular/hexagonal cell coverage cannot be assumed anymore. Here we assume the use of the modified Friis propagation model [3], and that values of the transmitter power, propagation exponent, and transmitter and receiver antenna gains are set at Pt = 2 dBW, g = 3, Gt = 17 dBi, and Gr = 9 dBi, respectively. The radio frequency bandwidth, noise figure, and frequency are brf = 3.5 MHz, Nf = 3 dB, and f = 3.5 GHz. The values of these parameters are extracted from the experimental fixed WiMAX demonstrator from [4]. In contrast with Mobile WiMAX, a limitation of IEEE 802.16-2004 is that it does not support sub-channelisation on the DL [5]. On the UL, the use of subchannelisation allows SS transmissions to only use 1/16 of the bandwidth assigned to transmissions from the BS, leading to a 12 dB link budget improvement [6]. The IEEE 802.16-2004 standard defines 16 sub-channels, where either 1, 2, 4, 8 or all sub-channels can be assigned to a SS. Each user in each different sub-channel may use a different MCS in successive bursts as long as he/she is using a different subchannel. Nevertheless, it is a worth noting that in the DL, because the MCS can be

368

F.J. Velez et al.

D+d

D

D+d R d

D D-d D-d

Fig. 9.1 Co-channel interference for the DL where the SS is at a distance d (0  d  R) from the centre of the cell (D is the reuse distance). The worst-case occurs when d = R

chosen at burst level, there is also the flexibility to in effect use different MCS by different users (even without sub-channelisation), as they may use consecutive bursts within a frame. In snap-shot simulations, averaging the generated interference by just placing all SSs in the center of the cell is not the correct procedure due to the non-linear influence of pathloss at different distances [7]. The contributions to interference from SSs equally distributed all over the cell surface area needs to be taken into account [7]. However, in this paper, in our chosen analytical approach for Fixed WiMAX, we do not follow this approach. Instead we consider the worst-case interference scenario, where the UL interferer is located at the edge of the neighbouring cell whilst considering that the user at the central cell can move between the centre and the edge of the cell.

9 System Capacity

369

With omnidirectional antennas, the worst-case for interference geometry corresponds to the case where the SS in Fig. 9.1 is at the cell edge, hence d = R. The carrier-to-interference ratio in the DL is given by Eq. (8.11). With tri-sectored cells (120 sectors), Eq. (8.11) becomes [8]: 1 C= ¼ I ðr þ 0:7Þg þ ðr  0:22Þg ; cc cc

(9.1)

which is valid for both links; rcc is given by (8.12). Note that in the omnidirectional case, the equation for the carrier-to-interference ratio in the UL results from the respective reuse geometry, where interferers are all at a distance D-0.866R from the central cell BS. This is given by: g

C= ¼ ðrcc  0:866Þ : I 6

(9.2)

These above equations only consider the first tier of co-channel interference: this assumption is generally only valid if a high value of the propagation exponent is used. If lower values for the propagation exponent are considered, interference to at least the second tier needs to also be considered [2]. To satisfy this, for example, for Eq. (8.11), terms proportional to 2(2rcc+1)g, 2rccg and 2(2rcc1)g need to be added to the denominator. The differences caused by the presence of sub-channelisation and sectorization can be interpreted by analysing the curves for CNIR as a function of the co-channel reuse factor, rcc, with R as a parameter. To produce these curves, the power of the carrier is obtained by computing the power received by an SS at a distance R from the BS, while the computation of the interference depends on the UL and DL configuration, and also on the use of sectorization. This can be computed for a fixed R by making the same considerations for frequency reuse as in Eqs. (8.11), (9.1) and (9.2). For the sake of simplicity, the modified Friis equation is used with different values of the propagation exponent, g, depending on the environment (g = 3 is considered in many numerical examples, as it may be suggested from the experimental work in a suburban area from [4]). The noise power is computed by using the following equation:   N½dBW  ¼ 204 þ 10  log brf ½Hz ;

(9.3)

where brf is the channel radio frequency bandwidth. In the sub-channelisation case, brf should be divided by 16. From these curves for the achievable CNIR as a function of rcc, by considering the values of CNIRmin, it is straightforward to obtain the maximum supported physical throughput at the cell edge (distance R) in a simplified way, that is, by considering that users are uniformly distributed on the cell but not considering the mixture of services and applications and the exact details for the corresponding

370

F.J. Velez et al.

“multiplexing” characteristics. For hexagonal-shaped cells, K = 1, 3, 4, and 7 correspond to reuse factors rcc = 1.732, 3.000, 3.464, and 4.583. We have utilised these values. The variation of CNIR with rcc without considering relays are presented in Chapter 8.

CNIR Versus Physical Throughput with Relays

9.3 9.3.1

Formulation

In this section the analysis of throughput is performed by considering the use of relays. In these topologies, a cell is composed by the central coverage area, served by the BS, and three 240 sector coverage areas, served by individual RSs (RS1, RS2 and RS3), Fig. 9.2. While the BS backhaul is assured in the usual terms for mobile communications (e.g., cable or micro-wave radio link), RS backhauling is supported by using special specific sub-frames within the radio channel created for this purpose [9]. The central coverage area BS may have omnidirectional or tri-sectored antenna. In the latter case, if Frequency Division Duplexing (FDD) is considered, more channels are needed which, in turn, allows for making extra resources available to the RSs (as separate frequency channels are made available at itch sector). While the BS backhaul is assured in the usual terms for mobile communications (e.g., cable or micro-wave radio link), RS backhauling is supported by using special specific sub-frames within the radio channel created for this purpose. Figure 9.3 shows the fixed WiMAX FDD mode frame structure assumed in this work. The WiMAX frame is divided into DL and UL sub-frames which, in turn, include sub-sections with the following purpose, Fig. 9.3: l l

BS to MS communications MS to BS communications

RS2

RS1 BS RS3

UL DL

Fig. 9.2 BS, RS and respective “hexagonal” coverage areas

MS

MS

9 System Capacity

371 1/3

2/9

2/9

2/9

DL (to RS2)

DL (to RS3)

DL Sub-frame DL (to MS)

DL (to RS1)

UL Sub-frame UL (from MS)

3/45

2/45

UL UL (from MS to BS) (from MS to RS1)

2/45

DL (from RS1 to MS)

2/45

UL UL (from MS (from MS to RS2) to RS3)

2/45

2/45

DL (from RS2 to MS)

DL (from RS3 to MS)

2/45

UL UL UL (from RS1 (from RS2 (from RS3 to BS) to BS) to BS)

Fig. 9.3 Frame structure for UL and DL sub-frames with relays (omnidirectional BS)

l l l l

BS to RS communications RS to BS communications RS to MS communications MS to RS communications

It is worthwhile to note that the UL sub-frame supports DL communications from RS to MS. As we assume asymmetrical communications between the UL and DL with an asymmetry factor of 1/5 as shown in Fig. 9.3, the UL sub-frame may make these extra resources available for DL RS communications. The advantage of using relays becomes from the fact that the co-channel interference now comes from cells at a longer distances because the real distance is given by: pffiffiffi D ¼ 3 kR

(9.4)

The corresponding cell geometry is presented in Fig. 9.4. The cell is formed by a central coverage zone with hexagonal shape and three hexagonal outer coverage areas with 240 sectored antennas, as shown in Fig. 9.4, each occupying two thirds of the area relatively to the central coverage zone. This different approach corresponds to consider three times the coverage area becomes [6, 7]: Amultihop ¼ 3:Asinglehop

(9.5)

Different cases are discussed for the DL and UL, for the communications from and to RS and BS.

372

F.J. Velez et al. BS position RS position

A multihop

= 3

3 2

3 R 2 = 3 * A singlehop

R

D

Fig. 9.4 Cell with RS at the edge of the central coverage area

9.3.2

Assumptions

A set of assumptions is considered in this research on frequency reuse for fixed WiMAX with relays with a frequency reuse pattern (or cluster size) K ¼ 3. For downlink (DL), the objective is to maximize the supported throughput. The optimization process is twofold [9]: l

l

l

l

The offered throughput from BS to RS needs to be maximized, that is, for a hexagonal coverage area with radius R the throughput Rb(R) needs to be the highest possible. The offered throughput to SSs needs to be maximized. It is further be divided into two points: Maximization of Rb-sup at the SSs on the central coverage area – By considering our assumptions for DL and UL frames (to cope with the Relay Station communications, Fig. 9.3) the DL throughput at the central coverage area is approximately 1/3 of the total one Maximization (Rb-sup)RS in SSs at the three relay coverage areas – The maximum throughput at the RS coverage area is Rbmax=min{Rb(R), (Rbsup)RS} multiplied by 2/9, that is, 2/9·Rbmax, where Rb(R) is the maximum throughput at

9 System Capacity

373

the edge of the central cell, at a distance R from the BS, and (Rb-sup)RS is the total throughput that may be supported at the RS coverage area if the Rs backhaul could support it (and considering the total frame duration) Relay station antennas for the communication with the BS are considered to be directional (e.g., 120 sectored ones); so that they only receive/cause interference from/at two BSs at distance D+R, as it will be shown in the formulation. From the communications at the relay, in our hypothesis for frames, it is only possible to achieve 2/9·Rbmax for the supported throughput at the whole RS coverage area. However, only if the BS to RS link supported throughput is enough the total throughput is guaranteed. In practice, the throughput at a distance d from the RS, Rb(d) depends on the supported modulation and coding scheme (MCS), and is given by [9]: 2 Rb ðdÞ ¼ Rb ðRÞ  AuxFactorðdÞ 9

(9.6)

where d is the distance to the RS and Rb(R) is the maximum throughput at the edge of the central cell, at a distance R from the BS and AuxFactor(d) is given in Table 9.1. Let’s assume as an example that the 16-QAM ½ MCS is supported in the central coverage area. Table 9.1 shows the values for AuxFactor(d) if the MCS ID that may be guaranteed for the CNIR(d) from the RS coverage area is 1, 2, 3, 4, 5, 6, 7 or 8. The 16-QAM ½ MCS is shown in bold in Table 9.1. In practice, the rule for RS is the following: If the MCS supported at a distance d from the RS is higher or equal the one supported in the BS-to-RS link (16-QAM ½ in this example) the throughput for RS will be 2/9 Rb(R); otherwise the throughput will be 2/9 (Rb-sup)RS. For UL, the maximization of supported throughput, Rb-sup is also twofold [9]: l

l

The supported throughput from RS to BS needs to be maximized. At the BS from RS it is only possible to achieve 2/45 Rb-sup The supported UL throughput needs to be maximized. It is further divided into two parts

Table 9.1 AuxFactor (d) for different values of the MCS ID for the communications to the SSs at RS coverage area ID MCS CNIRmin (dB) Physical thr. (Mbps) AuxFactor(d) 1 BPSK ½ 3.3 1.41 1.41/5.64 2 BPSK 3/4 5.5 2.12 2.12/5.64 3 QPSK ½ 6.5 2.82 2.82/5.64 4 QPSK 3/4 8.9 4.23 4.23/5.64 5 16-QAM ½ 12.2 5.64 1 6 16-QAM 3/4 15.0 8.47 1 7 64-QAM 2/3 19.8 11.29 1 8 64-QAM 3/4 21.0 12.27 1

374 l

l

F.J. Velez et al.

Maximization of the supported UL throughput at the RS (from MS) in the central coverage area (the maximum achievable throughput is 3/45 Rb-sup) Maximization of the offered throughput from SSs to RSs at the three RS coverage areas (the maximum is 2/45 (Rb-sup)SS)

Since relay station antennas are directional thus base station only receive interference from two BS at distance D+R. At the relays station it is only possible to achieve 2/45 (Rbsup)SS, where (Rbsup)SS refers to the supported throughput from SS to RS. This traffic will only reach BS if the RS to BS radio link supports such value of the throughput.

9.3.3

DL Scenarios

For the DL there are three different cases that need to be individually analyzed: 1. BS to SS: BS to SS communication is the simple case Fig. 9.5 is discussed before the same formula for C/I is used as discussed in (8.23). 2. BS to RS: In the case of BS to RS communication one assumes that RSs are using directional antennas of 120 Fig. 9.6 and only receive interference from

Fig. 9.5 DL scenario

9 System Capacity

375

Fig. 9.6 DL scenario with 120 RS sectorial antennas and 240 sector RS coverage area

two BSs, Fig. 9.8. This ultimately effects and enhances the C/I by a significant amount as shown in the later discussion. Therefore (D+aR)g/Rg=(rcc+a)g has a coefficient 2 while a = 1 and C/I is given by:   C 1 ¼ (9.7) I 2ðrcc þ 1Þg

3. RS to SS: In the case of RS to SS, the SS receives interference from four neighbouring RSs, Fig. 9.7. The distances between cell centres, RS and SS shown in Fig. 9.5 measured by using Autocad 2008 in a worst case situation where RS is at the edge of the coverage area. On the basis of measured distances the coefficients of R, in(D+aR)g, are calculated as given below [9]: pffiffiffi D ¼ 3R 3 ¼ 519:615 m

(9.8)

R ¼ 100 m

(9.9)

376

F.J. Velez et al.

Fig. 9.7 Distances from the RS interferers to the SS

There are two RS at 435.8899 m from the envisaged SS: 435:89  519:62 ¼ 0:837 100

(9.10)

There is one RS at 529.1503 m from the SS: 529:15  519:62 ¼ 0:0953 100

(9.11)

and one at a distance 608.27 m: 608:28  519:62 ¼ 0:8866 100

(9.12)

Hence, C/I is given by [9]: C Rg ¼ g I 2ðD  0:8372RÞ þ ðD þ 0:09535RÞg þ ðD þ 0:8866RÞg

(9.13)

9 System Capacity

9.3.4

377

UL Scenarios

For the UL there also are three different cases that need to be analyzed individually: 1. From SS to BS: In case of SS to BS there is interference from six surrounding SS. Thus carrier-to-interference ratio is given by

C ðrcc  0:866Þg ¼ I 6

(9.14)

2. From RS to BS: In this case we have assumed that RS antennas are 120 sectored one. Thus, the BS at the central cell only receives interference from two RS, at distance D+R, Figs. 9.8 and 9.9. The carrier-to-interference ratio is given by C ðrcc þ 1Þg ¼ I 2

(9.15)

3. From SS to RS: In this case RS receives interference from four SSs in neighbouring cells. By using the same procedure to measure the distances

Fig. 9.8 The decrease of the co-channel interference by using directional antennas at RS

378

F.J. Velez et al.

Fig. 9.9 Distances from RS to BS in the UL

between cell centres, RS and SS, the following values were obtained for the coefficients a of R: 0.8761, 0.82776, 0.80762. As a consequence, C/I is given by: C Rg ¼ I 2ðD  0:8761RÞg þ ðD  0:082776RÞg þ ðD  0:80762RÞg

9.4 9.4.1

(9.16)

Supported Physical Throughput Implicit Function Formulation

To guarantee Fixed WiMAX coverage with no coverage gaps near cell edges, the CNIR must be higher than 3.3 dB throughout the cell. This value corresponds to the minimum CNIR in order to use the BPSK ½ MCS (see Table 9.1). The assessment of the supported cell/sector physical throughput (per transceiver), Rb, as a function of the distance, d, produces a staircase-shaped curve

9 System Capacity

379 Rb

R

d

Fig. 9.10 Areas of the coverage rings where a given value of physical throughput is supported

indicating that higher maximum achievable throughputs are supported near the centre of the cell (see Fig. 9.10). As throughput is not constant over the whole coverage area for cellular planning proposes (where R is the cell radius), the supported throughput is obtained by computing the average supported throughput in the cell. As stated previously, in contrast to [6, 7], worst-case scenarios for interference geometry are considered here. There are J different coverage rings in each coverage zone, each supporting a different MCS (for instance, J = 4 in Fig. 9.10). The distances that correspond to the steps between consecutive MCS are represented by dj, j = 1, 2, . . ., J. Here we denote the order of the MCS as MCSj. The number of different coverage rings is given by: J ¼ MCS1st  MCSlast þ 1;

(9.17)

where MCS1st and MCSlast represent the MCS for the first and last coverage rings, respectively. If only one frequency channel is considered per cell, the supported throughput is obtained as [8, 10]:

380

F.J. Velez et al.

ÐÐ Rbsup ¼

O

  J  pffiffi  P 3 3 2 2  R  d  d ð Þ Rb ðd; R; K Þdxdy b MCS1st þ1j j j1 2 j¼1 pffiffi pffiffi ¼ ; 3 3 3 3 2 2 2 R 2 R

(9.18)

where the 2D integral is performed over the hexagonal shape of the cell. It is computed by weighting the supported physical throughput in each concentric coverage ring by the size of the ring where that value is supported. The contribution of each of the transmission modes is thus considered. MCS1st, MCS2nd, . . ., MCSJth can be obtained in the following way

  MCSj CNIR½dB ¼

8 > > > > > > > > > > > < > > > > > > > > > > > :

0; CNIR > 5:64; l ¼ 5 > > > 8:47; l ¼ 6 > > > > 11:23; l¼7 > : 12:27; l ¼ 8

(9.20)

CNIR(Rb) is not a bijective function. Therefore, the value of CNIR that corresponds to a given Rb is the minimum value of CNIR, that is, CNIRmin, that supports a given throughput Rb. Hence d0 = 0, and     dj ¼ cnir 1 min CNIR ðRb ÞMCS1st þ1j ; j ¼ 1; . . . ; J:

(9.21)

Figure 9.11 presents the correspondence between the CNIR vs. propagation distance curve and the stepwise function that represents the CNIRmin threshold for each MCS versus Rb. This Figure illustrates how the mapping between CNIR and

381

CNIR

CNIR

9 System Capacity

min CNIR MCS

J -2

min CNIR MCS

J -1

min CNIR MCS

dJ

d

Rb-MCS J -2

dJ-1

Rb-MCS J

dJ-2

Rb-MCS J -1

J

Rb

Fig. 9.11 Correspondence between the physical throughput for rings J, J1, J2, . . ., and the minimum CNIRs of consecutive MCS that map to step distances dJ, dJ1, dJ2, . . .

supported physical throughput relates to step distances between consecutive MCS dJ, dJ1, dJ2, . . . . This Figure illustrates how the mapping between CNIR and supported physical throughput relates to step distances between consecutive MCS dJ, dJ1, dJ2, . . . . In the context of the experimental work performed within our research group, results have fitted the modified Friis equation to some ranges of coverage distances in Fixed WiMAX [4]. According to the modified Friis equation, the received power is given by pr ðdÞ ¼

pt  gt  gr  l2 ð4pÞ2 d g

:

(9.22)

where 0  d  R, l is the wavelength, Pt, Gt and Gr (the latter ones are in dB), and g is the propagation exponent.

9.4.2

Without Relays

In the DL, for a given R, the reuse distance is given by D = rccR, and the interference at a distance d from the BS is computed by the following approximate equations iðd; D; RÞ ¼

p t gt gr l2 ð4pÞ2



 2 2 2 þ þ ; ðD  dÞg Dg ðD þ dÞg

(9.23)

382

F.J. Velez et al.

iðd; D; RÞ ¼

pt g t gr l2 ð4pÞ2



 1 1 þ : ðD  0:7  dÞg ðD  0:22dÞg

(9.24)

Equation (9.23) is applied in the omnidirectional BS antenna case while Eq. (9.24) is applied in the tri-sectored case. Under sectorization, only two interference sources need to be considered. Although these formulas are both valid for the DL (9.24) is also valid for the UL. For the omnidirectional case on the UL, Eq. (9.2) for interference can still be applied as it does not depend on d, as the distances from the SS interferes to the cell BS are D-R. Note that, as for Eqs. (8.11), (9.1) and (9.2), the second tier of interference would also need to be considered if lower values of the propagation exponent were used. We are aware that in this paper we do not consider per sub-channel equivalent SINR (or CNIR) computations when sub-channelisation is used. These computations could be performed accounting either for exponential effective SINR mapping (EESM [11–14]), effective code rate map (ECRM), or mean instantaneous capacity (MIC [6]). The consideration of these compression techniques may be needed in the presence of selective fading to adapt the curves to actual CNIRs in the UL. In the following, five different cases are addressed: l

l

l

l

l

The DL in the absence of sub-channelisation and sectorization (which we denote as the “DL – omnid”. case) The UL in the absence of sub-channelisation and sectorization (the “UL– omnid”. case) The UL in the presence of sub-channelisation and absence of sectorization (the “UL – sub-channel”. case) The UL and DL in the absence of sub-channelisation and the presence of sectorization (the “UL & DL – sector”. case) and The UL in the presence of sub-channelisation and sectorization (the “UL – subch. & sector”. case).

Figure 9.12 presents the curves for Rb(d) for K = 4 with a coverage distance R = 2,500 m. Without using sub-channelisation or sectorization, the DL performance is clearly better than the UL one. However, when sectorization is considered, higher physical throughputs are achievable. Besides, the better results are obtained when both sub-channelisation and sectorization are used. In this case, the highest physical throughput reaches 12.27 Mb/s, which is achieved for distances up to d  1,500 m. As an example, Fig. 9.13 presents a comparison of the results for the cell physical throughput for different reuse patterns K = 1, 3, 4 and 7, for “UL – subch. & sector”. and R = 2,500 m. If one carefully analyses the difference in the areas below the curves in these Figures, the following conclusions can be extracted: l

Although the physical throughput clearly increases with the use of sectorization only with the simultaneous use of sectorization and sub-channelisation in the UL the highest order MCSs are possible near the cell edge.

9 System Capacity

383

14 12

Rb [Mb/s]

10 8 6

DL - omnid. UL - omnid.

4

UL - subchannel.

2

UL & DL - sector. UL - subch. & sector.

0

0

500

1500

1000

2000

2500

d [m]

Fig. 9.12 Example of the variation of the physical throughput as a function of d for R = 2,500 m and K = 4

14 12

Rb [Mb/s]

10 8 6 4

K=7

2

K=3

K=4 K=1

0 0

500

1000

d [m]

1500

2000

2500

Fig. 9.13 Variation of the physical throughput in the “UL – subch. & sector.” case versus d, for R = 2,500 m

l l

With no improvement technique the DL performs better than the UL. A value for the reuse pattern K = 7 only presents a slight advantage relatively to the consideration of K = 4 or 3, whose behaviour is very similar; this is mainly true in the “DL” and “UL & DL – sector”. Cases, whose curves are not presented here. It should be noted here that K = 1 is not supported without the use of sectorization.

By applying Eq. (9.18) to the results for the cell physical throughput (as the ones from Figs. 9.12–9.13), one obtains the curves for the supported throughput as a

384

F.J. Velez et al. 5

Rb-sup [Mb/s]

4

3 2 UL&DL - sector. 1 0

UL - subch. & sector.

0

1000

2000

3000

4000

5000

4000

5000

R [m]

Fig. 9.14 Supported sector physical throughput versus R for K = 1

9 8

Rb-sup [Mb/s]

7 6 5 4 3 DL - omnid.

2

UL&DL - sector.

1 0

UL - subch. & sector.

0

1000

2000

3000 R [m]

Fig. 9.15 Supported cell/sector physical throughput versus R for K = 3

function of R for K = 1, 3, 4 and 7 presented in Figs. 9.14 through 9.17, respectively. Some of the curves with no sub-channelisation are either impossible to obtain at all or after a given R, for example, for K = 1, 3 or 4. This is because the physical throughput on the outer coverage ring of the cell reaches 0 Mb/s and full cell coverage may not be guaranteed. The supported throughput results for K = 4 are slightly worse than the ones for K = 7 but they are still acceptable (where there is the advantage under K = 4 of using only 4/7 = 57% of the spectrum bandwidth). For K = 3, although the degradation compared to K = 7 seems high compared with K = 4, only 3/4 = 75% of the bandwidth is used. This reduction in spectrum bandwidth is, however, not as much as between K = 7 and K = 4. Using K = 1 is

9 System Capacity

385

10

Rb-sup [Mb/s]

8

6

4 DL - om nid. UL - om nid. UL - subchannel. UL&DL - sector. UL - subch. & sector.

2

0

0

1000

2000

3000

4000

5000

R [m]

Fig. 9.16 Supported cell/sector physical throughput versus R for K = 4

12 10

Rb [Mb/s]

8 6 DL–om nid. UL– om nid. UL – subchannel. UL & DL – sector. UL - subch. & sector.

4 2 0

0

1000

2000

3000

4000

5000

R [m]

Fig. 9.17 Supported cell/sector physical throughput versus R for K = 7

advantageous because only a small portion of spectrum is needed. If tri-sectorization is used, as sub-channelisation is not supported in the DL, a fractional use of the WiMAX channels is not possible and three different channels are needed, one for each sector. In this case, the total supported throughput is three times the sector average throughput. With 3.5 MHz channels, only 10.5 MHz are needed for each link direction.

386

F.J. Velez et al.

Table 9.2 Average supported throughput for low coverage distances with different Ks with simple assumptions DL – omnid. UL – omnid. UL &DL – sector. UL – sector. & subch. Rbsup K¼1 K¼3 K¼4 K¼7

– 5.051 6.100 8.024

– – 4.623 6.985

4.515 8.160 9.149 10.689

4.591 8.241 9.255 10.813

100.0 80.0 Area [%]

1.41 2.12

60.0

2.82 4.23

40.0

5.64 8.47

20.0

11.29 12.27

250 500 750 1000 1250 1500 1750 2000 2250 2500 2750 3000 3250 3500 3750 4000 4250 4500 4750 5000

0.0

R [m]

Fig. 9.18 Area covered by each MCS versus R for K = 4, in the “UL – subch. & sector” case

It is however worthwhile to compare results for the supported throughput among different values of K by assuming, for the sake of simplicity, that only one single channel could be used (even with tri-sectorization). For short coverage distances, that is, up to 1,500–2,500 m, depending on the cases (the highest Rs occur with sub-channelisation), the average values of achievable supported throughput are presented in Table 9.2. When K decreases, if sectorization is not used the reduction in physical throughput is higher; that is, high values of throughput are only achievable through the use of sectorization. Moreover, for this range of coverage distances, the values of supported throughput are only 1–2% lower without sub-channelisation compared with the case where both sub-channelisation and sectorization are used. The comparison between the “UL – sector. & subch”. and “DL” cases shows a reduction in the supported throughput for the DL case of 38.7, 34.1, and 25.8%, for K ¼ 3, 4, and 7, respectively (note that for K ¼ 1, it is not possible to support users in the DL without sectorization). While with omnidirectional antennas there is a clear asymmetry between UL and DL traffic (see Figs. 9.16 and 9.17) it is evident that the UL and DL can be balanced through the use of sectorization. These curves can be better interpreted by analysing the variation with R of the cell area, in percentage, corresponding to each supported data rate, that is, for each MCS (according to Table 9.1). Figure 9.18 presents the corresponding curves for

9 System Capacity

387

100.0 80.0

1.41

Area [%]

2.12

60.0

2.82 4.23

40.0

5.64 8.47

20.0

11.29

4750

4250

3750

3250

2750

2250

1750

1250

750

12.27

250

0.0

R [m]

Fig. 9.19 Area covered by each MCS versus R for K = 7, in the “UL & DL – sector” case

K = 4 in the “UL – subch. & sector” case. From this, it can be observed that the highest values of the throughput are supported in the presence of sub-channelisation plus sectorisation, and correspond to the exclusive operation with the four highest order MCS for coverage distances of up to 4,750 m. Figure 9.19 presents the variation of the coverage area, in percentage, against R, for each MCS, but this time for the case K = 7 and “UL – subch. & sector”, which is presented as an example. By comparing these results with the ones for K = 4 one concludes that, for K = 7, the three highest order MCS are only supported up to R = 1,750 m, whereas, for K = 4, four different MCS are needed. However, for K = 7 and coverage distances higher than 2,500 m, the trend of enabling larger coverage distances while solely using the highest order MCS is not maintained anymore.

9.4.3

DL with Relays

The Fig. 9.20 shows the throughputs of the different scenarios of DL. 1. From BS to MS: Throughput from BS to MS is sufficiently high and it gradually decreases with the increase of the cell coverage distance R. In our assumptions, the frame structure is assigned 1/3 for DL. So, the DL throughput is obtained by the multiplying this factor 1/3 by the total obtained one. If we want to compare this value with the throughput in the RS coverage area, as they only have the coverage area, a normalized should be needed that is obtained by multiplying the throughput by this factor 2/3. 2. From BS to RS: The throughput from RS to BS remains high at constant level until 4 km and then suddenly decreases after 4 km. This throughput is obtained

388

F.J. Velez et al.

Supported Throughput [Mb/s]

3.5 3 2.5 2 1.5 from RS to MS from BS to RS from BS to MS from BS to MS normalized

1 0.5 0

0

1000

3000

2000

4000

5000

R [m]

Fig. 9.20 Results for the supported throughput as function of R in the DL with relays (K = 3, omnidirectional BS)

by using directional antennas at RS which greatly decreases the co-channel interference, Fig. 9.5. Only two RS receive interference from the central cell BS.

iðdÞ ¼

Pt Gt Gr l 2 ð4pÞ2

ðD þ dÞg

(9.25)

The receiver antenna gain Gr value was set to 28 dBi to obtain this throughput. 3. From RS to SS: The throughput for RS to SS is almost of same value as BS to SS. In our assumptions the frame structure is assigned 2/9 in case of RS to SS for DL. So, this is obtained by the multiplying this factor to the total obtained throughput.

9.4.4

UL with Relays and K = 3

Figure 9.21 show the results for throughput for different UL scenarios. 1. From SS to BS: Throughput for SS to BS resembles the previous scenarios of DL. Throughput decreases with the increasing cell coverage area R. In our assumptions, for SS to BS, the frame structure is assigned 3/45 for UL. So, the UL throughput is obtained by multiplying this factor 3/45 the total obtained one. If we want to compare this value with the throughput in the RS coverage area, as they only have the coverage area, a normalized should be needed that is obtained by multiplying the throughput by this factor 2/3.

9 System Capacity

389

Supported Throughput [Mb/s]

0.6 0.5 0.4 0.3 from MS to BS from RS to BS from MS to RS from MS to BS normalized

0.2 0.1 0 0

1000

2000

3000

4000

5000

R [m]

Fig. 9.21 Results for the supported throughput as function of R in the UL in the absence of subchannelisation (K = 3)

2. From RS to BS: The throughput from RS to BS remains high at constant level until 4 km and then suddenly decreases after 4 km. This throughput is obtained by the use of directional antennas at RS which greatly decreases the co-channel interference, Fig. 9.8. The receiver antenna gain Gr value is set to 28 dBi to obtain this throughput. 3. From SS to RS: Throughput for SS to RS is almost of same value as SS to BS. In our assumptions, for SS to RS, the frame structure is assigned 2/45 for UL. So, this is obtained by the multiplying this factor to the total obtained throughput.

9.4.5

Use of Sub-Channelisation

In the case of IEEE 802.16-2004 standard does not support the sub-channelisation for DL. However, it may be supported in the UL. Therefore, to obtain the sufficient level of throughput for UL the sub-channelisation is used. Throughput analysis for the UL scenario from the Fig. 9.22 is performed with sub-channelisation. The main change is the fact that the throughput from RS to BS now remains constant when the cell coverage area varies. However, there are no remarkable differences in the other curves. The throughput analysis from Fig. 9.11 for UL scenario is presented with subchannelisation. In the case of RS to BS throughput the main change is the fact that the throughput from RS to BS now remains constant when the cell coverage area increases. So, there is constant level of throughput at all distances of cell coverage area R. However, one can see from the results that there is no remarkable difference in the other curves.

390

F.J. Velez et al.

Supported Throughput [Mb/s]

0.6 0.5 0.4 0.3 from MS to BS from RS to BS from MS to RS from MS to BS normalized

0.2 0.1 0 0

1000

2000

3000

4000

5000

R [m]

Fig. 9.22 Results for the supported throughput as function of R in the UL with sub-channelisation (K = 3)

9.5

Throughput with Sectorization, Relays and K = 1

The analysis with K = 1 was tried with omnidirectional cells but the communication from MS to BS was impossible. As the topologies with RSs are more favourable, it is worthwhile to consider tri-sectored antennas at the BS of the central coverage area. We used Eq. (8.23) to compute the carrier-to-interference ratio from/to pffiffiffi the BS at the central cell (for DL and UL, respectively) but now with D ¼ 3 kR. The formulation for the communications between RSs and MSs is the same, as well as the one for the communications between RSs and BSs. By considering trisectored antennas we need to have one different channel (i.e., frequency carrier) for each sector. This way, more resources are made available to the RSs, and we can consider the assumptions from Fig. 9.23 for the DL and UL sub-frames. The asymmetry factor between the UL and DL is 3/7 in this case. Figures 9.24 and 9.25 show the results for the supported throughput for the DL and UL, respectively. Although the supported throughput per channel may be lower (e.g., 8.25 Mb/s at 750 m against 9.79 Mb/s in the omnidirectional case), the supported throughput from BS to SS (with K = 1, tri-sectorization) is clearly higher than the one for “K = 3, omnidrectional BS” (e.g., a normalized throughput of 3.30 Mb/s in tri-sectored case at 750 m against 2.02 Mb/s in the omnidirectional case). This is owing to the more favourable frame format (the asymmetry factor is 3/7 against 1/5 in the omnidirectional case). Nevertheless, with tri-sectorization and K = 1 it is a worth noting that the BS to SS communication is not the most limitative one anymore. In fact, the RS to SS link shows now the lowest throughput (e.g., 2.17 Mb/s for R = 750 m). These values are still slightly larger than the achievable ones in the omnidirectional central coverage area case with K = 3. Besides, it is worthwhile to note that, as there is one channel

9 System Capacity

391 3/5

2/5

DL Sub-frame DL (to MS)

9/35

6/35

DL (to BS)

6/35

2/5

UL Sub-frame UL UL (MS to BS) (MS to RS)

UL (RS to BS)

DL (RS to MS)

Fig. 9.23 Frame structure for UL and DL sub-frames with relays (tri-sectored BS)

Supported Throughput [Mb/s]

5

4

3

2 BS-to-MS BS-to-RS RS-to-MS BS-to-MS-normalized

1

0

0

1000

2000

3000

4000

5000

R [m]

Fig. 9.24 Results for the supported throughput as function of R in the DL with relays (K = 1, tri-sectored BS)

assigned to each sector the total cell throughput (to be fed into the cost/revenue optimization procedure) is achieved by multiplying this sector throughput by three. In the future, we also may analyse the presence of sub-channelization in the UL with K = 1, and its impact for the longest coverage distances. The case with trisectorized BS antenna and K = 3 is analyzed in [9].

9.6

Conclusions

In this work, a model to compute the supported physical throughput as a function of the achievable CNIR has been proposed for Fixed WiMAX. Frequency reuse topologies have been explored for 2D geometries that are commonly used in rural

392

F.J. Velez et al.

Supported Throughput [Mb/s]

2.5

2

1.5

1 BS-to-MS BS-to-RS RS-to-MS BS-to-MS-normalized

0.5

0

0

1000

2000

3000

4000

5000

R [m]

Fig. 9.25 Results for the supported throughput as function of R in the UL with relays (K = 1, tri-sectored BS)

and suburban environments, and the basic limits for system capacity and cost/ revenue optimisation have been obtained by considering simple assumptions. It is assumed that line of sight propagation to the bases station is achieved in a high percentage of the cell, reducing the impact of selective fading, through allowing dimensioning to be done by GIS cellular planning tools. For a given coverage area, throughput is a stepwise function that decreases as the distance to the base station increases. Its value depends on the supported MCS for each coverage ring. In this chapter, the supported throughput has been computed by weighting the available throughput at each coverage ring with the area (or size) of the ring. Throughput typically decreases as the cell radius increases, although through the use of sub-channelisation it is possible to keep its value steady at least up to a cell radius of 5,000 m. With the use of sectorization, the supported throughput is higher, corresponding to the use of the highest order MCSs. However, as sectorised equipment is more expensive and there is a need for three times more bandwidth, costs are also higher. In this Chapter formulations were also proposed to account for the interference in cellular coverage and reuse geometries without and with the use of relay. In our proposal for relays, the FDD mode is considered and the frames need to guarantee resources for BS-to-MS communications but also for BS-to-RS and RS-to-MS communications. These requirements leads to a 1/5 asymmetry factor between the UL and DL in the omnidirectional BS case and to a 3/7 asymmetry factor in the case of tri-sectored BSs. pffiffiffi Although the reuse distance is augmented by a factor of 3, it was first shown that the use of relays corresponds to lower values of the supported throughput for K = 3. The presence of sub-channelisation only improves the results for the highest values of R.

9 System Capacity

393

The consideration of tri-sectored BS antennas with K = 1 (whilst keeping the number of required channels – equal to 3) did not enable to obtain values of the throughput comparable to the ones without using relays, although frame format is more favourable. Relays can be cheaper than BS with full functionalities. As the use of relays may lead to lower costs it is worthwhile to analyse the impact of using them on costs and revenues.

References 1. G. Bauer, R. Bose, R. Jakoby, Three-dimensional interference investigations for LMDS networks using an urban database. IEEE Trans. Antennas Propag. 53(8), 2464–2470 (Aug 2005) 2. F.J. Velez, L.M. Correia, J.M. Bra´zio, Frequency reuse and system capacity in Mobile Broadband Systems: comparison between the 40 and 60 GHz bands. Wireless Pers. Commun. 19(1), 1–24 (Aug 2001) 3. T.S. Rappaport, Wireless Communications: Principles and Practice (Prentice Hall, Upper Saddle River, NJ, 2002) 4. P. Sebastia˜o, F. Velez, R. Costa, D. Robalo, A. Rodrigues, Planning and deployment of WiMAX networks. WIRE – Wireless Pers. Commun. (Aug 2009). doi: 10.1007/s11277009-9803-3 5. Hui Liu, Guoqing Li, OFDM-based Broadband Wireless Networks – Design and Optimization (Wiley, Hoboken, NJ, 2005) 6. J.G. Andrews, A. Ghosh, R. Muhamed, Fundamentals of WiMAX – Understanding Broadband Wireless Networking (Prentice Hall, Upper Saddle River, NJ, 2007) 7. C. Hoymann, S. Goebbels, Dimensioning cellular WiMAX part I: singlehop networks, in Proceedings of EW’2007 – European Wireless 2007 (Paris, France, Apr 2007) 8. F.J. Velez, V. Carvalho, D. Santos, R.P. Marcos, R. Costa, P. Sebastia˜o, A. Rodrigues, Aspects of cellular planning for emergency and safety services in mobile WiMax networks, in Proceedings of ISWPC’ 2006 – 1st International Symposium on Wireless Pervasive Computing 2006, Phuket, Thailand, Jan 2006 9. F.J. Velez, M.K. Nazir, A.H. Aghvami, O. Holland, D. Robalo, Cost/revenue Trade-off in the optimization of Fixed WiMAX Deployment with Relays, submitted to IEEE Transactions on Vehicular Technology (Dec 2009) 10. F.J. Velez, A.H. Aghvami, O. Holland, Basic Limits for Fixed WiMAX Optimization Based in Economic Aspects, accepted for publication in IET Communications – Special Issue on WiMAX Integrated Communications (Mar 2009) 11. R. Jain, Chakchai So-In, A.-K. Al Tamimi, System-level modeling of IEEE 802.16e mobile WiMAX networks: key issues. IEEE Wireless Commun. 15(5) (Oct 2008) 12. Sergey N. Moiseev, Stanislav A. Filin, and Mikhail S. Kondakov, Analysis of the Statistical Properties of the Interference in the IEEE 802.16 OFDMA Network, IEEE Wireless Communications and Networking Conference (WCNC 2006), vol. 4, pp. 1830–1835, Apr 2006 13. M.S. Kondakov, A.V. Garmonov, Do Hyon Yim, Jaeho Lee, Sunny Chang, Yun Sang Park, Analysis of the statistical properties of the interference in the IEEE 802.16 OFDMA Network, in Proceedings of WCNC 2006 – IEEE Wireless Communications and Networking Conference, Las Vegas, Nevada, USA, Apr 2006 14. 3GPP, 3rd Generation Partnership Project; Technical Specification Group Radio Access Network; Feasibility Study for OFDM for UTRAN Enhancement (Release 6), 3GPP TR 25.892 V2.0.0, 2004-06

Chapter 10

Business Models and Cost/Revenue Optimization Fernando J. Velez, M. Kashif Nazir, A. Hamid Aghvami, Oliver Holland, and Daniel Robalo

Abstract This Chapter starts by covering general aspects about the business models for WiMAX and then addresses the cost/revenue optimization for these networks, for cellular configurations without and with relays. In Fixed WiMAX, radio and network planning can be optimised by tuning a cost/revenue function which incorporates de the cost of building and maintaining the infrastructure and the effect of the available resources on revenues. From the cost-benefit analysis, one conclusion of this work is that given today’s hypothesis of price per MByte of information transfer of somewhere between 0.0025 € and 0.010 €, it is clear that, without considering the use of relays, the choice of reuse patterns 3 or 4 with sectorial cells is preferable to the use of omnidirectional cells with reuse pattern, K, of 7, as three times more resources are available in each cell. Besides, in nowadays networks, if there is a need for sparse BS deployments whilst reducing costs, K ¼ 1 may be a solution, as it presents higher profit for the longest coverage distances. In future networks, when costs will be lower, the advantage of sectorization is kept and will drive the deployment of tri-sectorization forward. Nevertheless, in this case K ¼ 1 will not be advantageous with tri-sectorization for the longest coverage distances anymore. This study also concludes that cell radii in the range 1000–1500 m is preferable, corresponding to profit in percentage terms of near the achievable maximum, while keeping costs acceptable. The WiMAX cost-benefit optimization is also explored for the case where relays are used to help on improving coverage while mitigating the interference. Results show that the use of relays with no sectorization in the BS leads to a lower profit (K ¼ 3). Also the use of sectorization (an example is presented for K ¼ 1) does not seem to enable larger profit. The optimum (maximum) values occur for coverage distances up to 1,000 m. In the DL, when the price per MB, R144, increases from 0.0025 €/min to 0.005 €/min the profit increases more than 100%. F.J. Velez (*) Instituto de Telecomunicac¸o˜es-DEM, Universidade da Beira Interior, Calc¸ada Fonte do Lameiro, 6201-001 Covilha˜, Portugal e-mail: [email protected]

R. Prasad and F.J. Velez, WiMAX Networks, DOI 10.1007/978-90-481-8752-2_10, # Springer ScienceþBusiness Media B.V. 2010

395

396

10.1

F.J. Velez et al.

Introduction

Although the oldest wireless Internet providers have been providing narrowband since perhaps 1992, the broadband wireless access (BWA) business is a relatively new phenomenon. Some early wireless broadband technologies like LMDS and MMDS tried to establish their roots and to become popular. These were fragmented and competing proprietary standards and did not prove to be 100% successful. They also lacked the broad industrial support like WiMAX is having through the WiMAX Forum. Along with WiMAX, the other nowadays technologies providing wireless broadband are HSDPA, HSUPA, and EV-DO etc. WiMAX deployment optimisation can be achieved by appropriately parameterising a merit function, taking costs and revenues into account. The optimisation of the cost/revenue trade-off provides a means of combining several contributing factors in cellular planning: determination of the reuse pattern, coverage distance, and the resulting supported physical throughput. Given the current state of national frequency spectrum assignments throughout Europe, the Fixed WiMAX achievable frequency reuse pattern, K, determines the reduction in the initial fixed cost if the required spectrum bandwidth is reduced to values comparable to the ones for Wideband Code Division Multiple Access (WCDMA) systems. In turn, the supported throughput will determine the achievable revenue, which has interdependencies with the use (or not) of sub-channelisation and/or sectorization. The optimization of the cost/revenue trade-off for different topologies is thus of fundamental importance, and can be achieved by varying system parameters and implied coverage and reuse distances. A cost/revenue function has to be developed by taking into account the cost of building and maintaining the infrastructure, and the way the number of channels available in each cell affects operators’ and service providers’ revenues. Fixed costs for licensing and spectrum bandwidth auctions (often known as “beauty contests”) should also be taken into account. The economic analysis is referred as a cost/ revenue performance analysis, because the optimisation (i.e., minimization) of cost does not necessarily mean the optimization of net revenues. Although one considers a project duration of 5 years as a working hypothesis in radio and network planning, it is decided in this paper to analyse costs and revenues on an annual basis. Furthermore, our analysis is under the assumption of a null discount rate. By no means is it intended to perform a complete economic study in this paper, the aim is simply to present initial contributions that facilitate cellular planning optimisation. Appropriate refinements would be needed to perform a complete economic analysis based on discounted cash flows (e.g., to compute the net present value). Besides addressing business models and plans in Sections 10.2 through 10.4 the main contributions from this chapter are the following. Section 10.5 describes the range of services offered. Section 10.6 addresses cost/revenue optimization, presenting the models without and with considering relays, as well cost/revenue optimization results and profit.

10

Business Models and Cost/Revenue Optimization

10.2

397

Business Model Definition

The term “business model” describes the interrelationship between different entities of the value network. A technology can be made successful only if it is supported by a successful business case. A complete business model comprises inputs from two domains, technical and economic. As shown in Fig. 10.1, a business model forms the link between creative technological ideas and the economic implementation of an innovation. A business model draws on a multitude of business subjects, including economics, entrepreneurship, finance, marketing, operations, and strategy. The business model itself is an important determinant of the profits to be made from an innovation. Sometimes it is seen that a mediocre innovation with a great business case may be more profitable than a great innovation with a mediocre business case. The basic business model is used to analyse the various service, organisational, technology and financial aspects. A right business model describes the interrelationships between different entities of the whole value network and the processes that take place between each of them.

10.3

Broadband Communications Business

There are many market drivers which has made broadband communications so popular. Some of them are new technologies, new applications, increased computing power and storage, mass production, price reductions, the Internet revolution and the competition. Also new technologies, the mass production of network components and low transmission costs are continuously creating new applications. At the same time, an extraordinary expansion of the Internet has occurred. It seems that it is not a killer application for the broadband market, but that Internet is a ‘killer network’, with a “killer” cocktail of applications and services. WiMAX is a highly versatile BWA technology platform and has relevance to a wide variety of application scenarios. The promise of WiMAX lays in its open

Technical Inputs

Fig. 10.1 Generic business model

Business Model Model

Economic Outputs

398

F.J. Velez et al.

standard design and capacity for interoperability. As a standard-based technology, with a huge ecosystem of developers and industry players, WiMAX is a technology of choice for the wireless broadband future. WiMAX wireless broadband is expected to evolve at a rapid pace, connecting the unconnected and bringing new possibilities to consumers and enterprises. With an OFDMA air interface, all IP architecture, adaptive modulation and coding techniques, and multiple advanced antenna options (including MIMO and beamforming), WiMAX promises high performance connections to deliver fixed broadband with deep penetration as well as high throughput. IEEE 802.16e, IEEE 802.16d and then IEEE 802.16m are expected to address a new breed of devices, applications and services that rely on broadband connections for enhanced user experience. Currently, there are very few commercial WiMAX compliant networks deployed. At the end of 2007 there were a total 181 operators globally. This number is expected to rise to 538 by 2012 [1]. New players are coming up in the WiMAX business. These are from diverse industries such as broadcasting, content studios, utilities and consumer electronics focused on providing fixed and mobile broadband mix services as an alternative to wireline technologies. The business model for the WiMAX deployment must consider all the aspects of design, deployment, and integration from the core network through the systems architecture, service edge, access network and devices.

10.4

Developing a WiMAX Business Plan

10.4.1 Market Research: Gathering the Input Parameters for a WiMAX Business Case In order to make a successful business case extensive market research has to be done. The inputs gathered from the research vary from country to country, but will help to deploy a feasible network deployment depending on the financial, technical and business parameters.

10.4.1.1

Target Market Characteristics

The first step in the market research is to decide the geographic area to be covered under the WiMAX services. This is measured in square kilometres of coverage for urban, suburban and rural geographies. The accurate network planning is possible when the terrain type of the geographical area is known. The terrain type can be classified under flat, moderate hilly or hilly. Geographic Information System (GIS) tools are used to explore aspects in network planning.

10

Business Models and Cost/Revenue Optimization

10.4.1.2

399

Spectrum

Licensed radio spectrum is the prime asset of a WiMAX network, and its value is determined by many factors, including its frequency, capacity and whether it is fixed or mobile. Recognizing the virtues of WiMAX for broadband wireless access the global regulators have allocated new spectrum channels. Globally the spectrum allocated for WiMAX is around 2.5 and 3.5 GHz bands. Around 500 companies worldwide have BWA licenses in the 3,400–3,600 GHz frequency bands while a few have licenses in the 2,500–2,690 GHz band .The typical deployment characteristics of the BWA bands are listed in Table 10.1.

10.4.1.3

Technological Parameters

Once the target market characteristics are found out and the bandwidth is known the technological parameters need to be decided. Nowadays, the first technological selection is between 802.16d or 802.16e. To achieve ubiquitous coverage throughout the entire geographical area, technical parameters such as link budget, spectral efficiency and antenna configuration need to be decided. These parameters along with the frequency to be used will help to calculate the coverage area (with no considerable interference) per cell site, hence, the total number of base stations to cover the desired geography. The radio characteristics of the WiMAX equipment selected for the analysis in this Section are listed in Table 10.2.

Table 10.1 Characteristics of 3.5 and 2.5 GHz frequency bands Features 3.5GHz 2.5GHz Total available spectrum About 200 MHz between 195 MHz, including guard –bands and 3.4 GHz and 3.8 GHz the MDS channels ,between 2.495 (varies from country to and 2.690 GHz country) Serviced offered Fixed, may allow mobile Fixed and Mobile FDD or TDD Mixed, some countries TDD/FDD specify FDD only while others allow either FDD or TDD Spectrum per license Varies from 2*5 MHz to 22.5 MHz per license, a 16.6 MHz 2*56 MHz block paired with a 6 MHz block. Total eight licenses License aggregation No Yes Allocation Worldwide except in US U.S., Canada, Latin America, Australia, expected in Asia

400

F.J. Velez et al.

Table 10.2 Radio characteristics Attribute Duplexing Channel bandwidth Adaptive modulation System gain Path loss for indoor CPEs Propagation conditions

10.4.1.4

3.5 GHz 2.5 GHz FDD TDD 2  3.5 MHz 5 MHz BPSK, QPSK, 16QAM, 64QAM 158 dB 157 dB 15 dB Urban, suburban and rural under NLoS conditions

Determining Financial and Capital Expenditure (CapEx)

There are certain financial parameters which need to assumed or taken into account before the operator starts the actual network planning. Critical evaluation regarding interest rates on borrowing and the expected returns need to be forecasted. The tax rate on profits, depreciation and the amortization period should also be taken into account. As in any major network deployment, huge initial capital is required to procure the WiMAX equipment as well as to acquire or lease sites. The setup of the access network also adds to the capital expenses (CapEx). This includes not only the base station costs but also the other components in the access network.

10.4.2 Market Planning 10.4.2.1

Competitive Analysis

There are a host of broadband service providers in the market today. The operator must take into account the penetration, type of services and their pricing levels by the existing broadband players. These include DSL and cable providers, on the wire-line side, and the 3G service providers, on the wireless side. A complete and understanding of the competitive environment for the broadband services in the operators market is necessary to develop a viable business plan.

10.4.2.2

Coverage Requirements

The main goal of the WiMAX operator is always to achieve the ubiquitous coverage throughout the entire metropolitan target area. An operator can develop a plan for rolling out the network based on different scenarios, for example, to consider the number of customers.

10

Business Models and Cost/Revenue Optimization

10.4.2.3

401

Capacity Requirements

Traditionally cellular networks were deployed to achieve ubiquitous coverage, with very little consideration for capacity requirements. This was the most reasonable approach as the only service offered was voice. More base stations were installed when the customer base and the number of services offered grew. However, with broadband, a range of services with varying Quality of Service (QoS) and Quality of Experience (QoE) are offered. To meet the customer expectations for these type of services it is of prime importance to predetermine the capacity requirements and accordingly deploy at the outset. This will ensure a quality user experience (even in the busiest ours in the highest density urban areas). Data density, expressed in Mbps per km2, is used to describe capacity requirements. Determining the data density requirements for a specific demographic region is multi-step process shown in Fig. 10.2. The service provider generally offer plans with varied service level agreements (SLAs) to appeal wide range of anticipated customer types in the target market segment. An SLA is a negotiated agreement between a service provider and a consumer for a subscription choice which typically defines the service level offered for a given monthly fee. There will be different sets of SLAs for residential and business segments. Depending upon the capacity to be offered the operator also need to decide on the oversubscription rate. Oversubscription means assigning a total committed information rate to a given base station that is greater than the base station capacity. This number is used to determine the peak busy hour traffic. As the number of users increases, a calculation using the oversubscription rate will tell the operator when to add new cells for the increased usage.

Market demand and forecast

Determine the target market segment

Area demographics

Services to be offered

Expected number of customers

Expected market take rate

Geographical Information System (GIS)

Fig. 10.2 Determining market driven capacity requirements

Required data rate (Mbps)

402

F.J. Velez et al.

10.4.2.4

Penetration

When building a broadband wireless access network, it is difficult (as well as challenging) to forecast number of subscribers that an operator can expect to sign over the life of the network. It generally takes a period of time for consumers to buy a new technology, a new service or opt for a new provider of that service. For some consumers the technology, service and provider have to be well-tested before they will sign up for the service. But the success of Wi-Fi will help WiMAX for a general acceptance of broadband wireless access. The rates charged for services by the operator will also have a marked effect on how quickly the technology and services will be adopted. Operators will also have a quicker adoption rate in past non-covered area than in areas that are currently well served. In Fig. 10.3, curves for the number of subscribers over a 10 year time frame are plotted. It shows that for first couple of years there are very few subscribers [2]. As the operator is deploying the network and expanding coverage the number of subscribers increases. Once the network is fully deployed, there will be an increase in subscribers until some level of saturation is reached.

10.4.2.5

CapEx

In analysing the business case, capital expenditure (CapEx) is calculated by taking into account the end to end network, as described in Fig. 10.4. The major CapEx components are the following: l l l

Spectrum licenses Customer Premise Equipment (CPE) Base station infrastructure 16 Percentage Penetration

14 12 10 8 6 4 2

10 ar Ye

ar

9

8 Ye

ar

6

7 Ye

Ye ar

ar

5 Ye

ar

r4

Fig. 10.3 WiMAX penetration curve

Ye

Ye a

2

ar 3 Ye

ar

1 Ye

Ye ar

Ye

ar

0

0

10

Business Models and Cost/Revenue Optimization

403

PSTN Fiber Network Businessl

Global Internet

Residential Market

Access Coverage Radio planning Capacity

BTS

Backhaul

WiMAX BS equipment WiMAX service edge network Wireless backhaul equipment Edge and core IP Wireline backhaul Installation & leasing

Edge

Core

Edge and core IP Networking elements Content management and delivery Media gateways

Fig. 10.4 WiMAX end-to-end network l l l

Edge, core and central office Wireless backhaul equipment Site development and acquisition cost

The CapEx is different for the incumbent compared to a new operator. For an incumbent wire-line operator, at least some part of the edge, core and central office equipment is ready. The only need is to add capacity to support the additional anticipated customers that would be covered by the WiMAX portion of the access network. Nevertheless, for the new operator large investment is required for edge, core and central office. As shown Fig. 10.4 the wireless portion the network begins at the fiber node or microwave link with a WiMAX BS or a wireless point-to-point link to a remotely located WiMAX BS. This backhaul link can be a WiMAX compliant point-to-point solution or some other point-to-point radio technology evolution. In addition to the WIMAX equipment, the base station may also include additional hardware such as the uninterruptible power supplies, electronic cabinets and some other additional equipment for the interface with the backhaul link. The costs for the civil works including antenna masts, conduits, cables, and the overall base station preparation also need to be considered. The base station infrastructure CapEx items are fixed in costs and are labour intensive. The cost is different for developing countries compared to developed ones. The civil works for urban area deployments are also higher compared to suburban and rural areas. 10.4.2.6

OpEx

As in any major network deployment, the total cost of ownership of a WiMAX network will comprise of operating expenses along with the capital expenditures.

404

F.J. Velez et al.

Table 10.3 Operating expenses

Operating expenses (OpEx) Site rental Site utilities Backhaul installation Backhaul cost Network maintenance Support and warranty Billing/CRM Marketing and advertisement Equipment maintenance Bad debt and churn General and administrative expenses

The operating cost comprises of the ongoing operational expense of managing and maintaining the WiMAX service. The major operational expenses are given in the Table 10.3. While considering the WiMAX business case its important for the operator to consider the various type of subscriber’s stations or terminals used with various SLAs. CPEs can be fixed (indoor and outdoor), WiMAX cards integrated in laptops, portable CPEs or embedded in consumer electronics (CE) devices. The business case should include the price of the CPE, it can be either subsidised by the operator or paid by the consumer.

10.5

Range of Services Offered

Today the broadband service no longer means just high speed Internet access. Broadband has successfully evolved to become the enabler of a bouquet of IP services. It’s no longer enough to provide the high speed internet facilitates such as surfing, e-mailing, file sharing, instant messaging, operators need to launch additional applications, too Therefore, an extensive review of service provider strategies must be made. WiMAX has to be more about services than it is about technology. The more services an operator is able to successfully leverage on its network, the more valuable the network becomes. The first step in defining the market is the range of services that would be offered to the potential customer. The right mix of services will allow the operator and the investors to realize swift return on their investment, generate strong revenue and grow market share. The portfolio of these services must address the various needs across the operators’ end user segments including basic residential, high speed residential, small to medium business, enterprise and government.

10.5.1 Classification of Services The classification of services and applications can be segmented into three main groups (Table 10.4):

10

Business Models and Cost/Revenue Optimization

Table 10.4 Broadband services Classification of services Content services

Information retrieval (browsing, surfing) Peer to peer and person to person

l l l

405

Applications Purchase of movies and music Leasing of movies and music Cultural services and entertainment: Video on demand(VoD) Events on demand TV channels and subscriptions On demand: news, sports, health, life style programs, etc. Substitutions: Performances like theatre, concerts, opera, cinema, etc. Lotteries and gambling: Online gambling Online betting Video transmission of the gambling event Gambling on automates Books, newspapers ,newsletters, journals: Books Online newspapers Newsletters Journals E-learning: Online education Online games Down loading and updating of gaming software Online games Traditional free surfing Information storage (film, photos, other information) Video conference Exchange of personal content Exchange of downloaded content Surveillance at home for elderly/sick Different home services

Content services Information retrieval and storage Person to person and peer to peer

10.5.2 Varying Terrain Conditions The calculation of the coverage area and of the performance for WiMAX networks have to be based on the terrain conditions of the intended service area. 1. Dense Urban: This is the city centre where many of the businesses are located as well as high density multiple dwelling residential units. These areas represent a challenging propagation environment due to the multipath caused by the multistorey buildings.

406

F.J. Velez et al.

2. Urban: Immediately surrounding the city centre would be more businesses and moderate to high density multiple dwelling units. Average building heights may be lower but the propagation environment will be equally challenging. 3. Suburban: This describes areas with lower density housing, primarily single family dwellings, and fewer businesses. Average building heights are lower and, on average, structures are more spread out, thus creating a more favorable propagation environment. 4. Rural: Moving further from the city centre, homes are further apart resulting insignificantly lower population density with scattered small businesses.

10.6

Cost/Revenue Optimization

10.6.1 Models The economics of cellular systems can be viewed from the points of view of the different entities: subscribers, network operators, service providers, the regulator, and equipment vendors [4–6]. In this paper, although it is possible that for mobile multimedia networks the network operator and service providers can be different entities, we do not distinguish them. Thus we are considering the operator/service provider’s point of view, whose primary bottom line is to improve his business. In the cellular planning process, the objective of the operator is to determine an optimal operating point that maximizes expected revenues. Examples of major decisions affecting this include the type of technology to be used, the size of the cell, and the number of radio resources in use in each cell. It is important to identify the main components of the system’s cost and revenues, in particular those that bear a direct relationship to either the maximum cell coverage distance or the reuse pattern. Here we consider the cost per unit area of a 2D system incurred during the system lifetime. The system is considered to have a transmission structure formed by a set of frequency carriers or channels (or the corresponding WiMAX sub-channels), each supporting a TDM frame structure. Each base station comprises a number of transceivers equal to the number of carriers assigned to the BS (or to the BS sector), which is assumed to be one in this study. That is, it is assumed as a simplification that one carrier will be sufficient per cell/sector. System cost has two major parts: (a) capital costs (normal backhaul, cell site planning and installation), and (b) operating expenses (operation, administration and maintenance) [7, 8]. The capital cost is taken to consist of: l l

A fixed part (e.g., licensing and spectrum auctions or fees) A part proportional to the number of BSs per kilometre or square kilometre (e.g., the installation costs of BSs including the cost of obtaining cell sites, the normal backhaul, and the cost of hardware and core equipment common to all), and

10 l

Business Models and Cost/Revenue Optimization

407

A part proportional to the total number of transceivers per kilometre or square kilometre (e.g., the cost of the transceivers)

It is assumed that the cost of the connection between BSs and the Switching Centre, that is, the fixed part of the network (e.g., the cost of laying fibre), is not a fixed cost. Instead, we consider this to be proportional to the number of BSs, which can be true if, for example, the mobile operator’s service is contracted from a fixed network operator. The operating cost during a system’s lifetime is taken to contain l l

A part proportional to the number of BSs per kilometre or square kilometre, A part proportional to the number of transceivers per kilometre or square kilometre.

These costs will be incurred on an annual basis. A similar approach was followed in [3] for hierarchical WiMAX–Wi-Fi networks. However, here we follow the approach from [9]. The cost per unit area is given by: C½=km2  ¼ Cfi½
=km2  þ Cb  Ncell=km2 :

(10.1)

where Cfi is the fixed term of the costs, and Cb is the cost per BS assuming that only one transceiver is used per cell/sector. The number of cells per unit area is given by: Ncell=km2 ¼

2 pffiffiffi ; 3  3  R2

(10.2)

and the cost per BS is given by [9]: Cb ¼

CBS þ Cbh þ CInst þ CM&O ; Nyear

(10.3)

where Nyear is the project’s lifetime (assumed here to be Nyear = 5), CBS is the cost of the BS, Cbh is the cost for the normal backhaul, CInst is the cost of the installation of the BS, and CM&O is the cost of operation and maintenance. The revenue per cell per year, (Rv)cell, can be obtained as a function of the supported throughput per BS or sector (in the omnidirectional and sectorial cases, respectively), Rb-sup[kb/s], and the revenue of a channel with a data rate Rb[kbps], RRb[€/min], by: ðRv Þcell ¼

Nsec  Rbsup½kb=s  Tbh  RRb ½
=min ; Rbch½kb=s

(10.4)

where Nsec is the number of sectors (one or three) Tbh is the equivalent duration of busy hours per day, and Rb-ch is the bit rate of the basic “channel”. In the

408

F.J. Velez et al.

tri-sectorial case, as one assumes that each sector has one different transceiver, there is a separate frequency channel available for it. The revenue per unit area per year, Rv[€/km]2, is obtained by multiplying the revenue per cell by the number of cells per unit area: Rv½
=km2  ¼ Ncell=km2  ðRv Þcell ¼ Ncell=km2 

Nsec  Rbsup½kbps  Tbh  RRb ½
=min : Rbch½kbps

(10.5)

The (absolute) profit is given by P½
=km2  ¼ Rv  C;

(10.6)

from which, the profit in percentage terms is given by: P½% ¼

10.6.2

Rv  C  100: C

(10.7)

Hypothesis Without Relays

Following the approach form [9], it is hypothesised that project duration is of 5 years and there is a null discount rate; costs and revenues are taken on an annual basis. We consider 6 busy hours per day, 240 busy days per year [10], and a revenue/price of a 144 kb/s “channel” per minute (approximately corresponding to the price of 1 MByte, as 144  60 = 8.640 kb  1 MByte), R144[€/min]. The revenue per cell can be obtained as: ðRv Þcell½
 ¼

Nsec  Rbsup½kbps  60  6  240  R144½
=min : 144½kbps

(10.8)

Diverse assumptions for the price of the 144 kb/s channel (or a MByte of information) are considered for each scenario. Two hypotheses are made for cost, denoted as A and B, as shown in Table 10.5. Hypothesis A is today’s situation. In the future, equipment prices will get lower with mass production, and spectrum bandwith prices will also reduce, thereby making Fixed WiMAX systems more accessible. This future case is hypothesis B. Assuming that the annual cost of a license is 50,000,000 € for 2  24.5 MHz bandwidth (UL & DL, K = 7), considering a total area of 91,391.5 km2 as the area of Portugal, for example, the fixed cost per unit area is Cfi ½
=km2  ¼

50;000;000 ¼ 108:24  110
/km2 : 91; 391:5  5

(10.9)

10

Business Models and Cost/Revenue Optimization

Table 10.5 Fixed WiMAX cost assumptions Costs Omnidirectional A B Cfi (€/km2) K=1 15.71 15.71 K=3 47.14 47.14 K=4 62.86 62.86 K=7 110.00 110.00 18,000 9,000 CBS (€) 10,000 1,000 CInst (€) 5,000 2,500 Cbh (€) 4,000 1,000 CM&O (€/year) Table 10.6 Required spectrum bandwidth for different cell configurations and reuse patterns

K 1 3 4 7

409

A

Tri-sectored B

47.14 141.43 188.57 330.00 30,000 18,000 5,000 6,000 BW (MHz) Omnidirectional 3.5 10.5 14.0 24.5

47.14 141.43 188.57 330.00 15,000 1,500 2,500 1,500

Tri-sectored 10.5 31.5 42.0 73.5

If one considers that only one carrier will be allocated to each cell (or sector), if K = 4 or K = 3 then the available BW (and the respective cost) will be 4/7 or 3/7 of the value for K = 7, respectively. Given that the total bandwidth, BW, is given by: BWomni½MHz ¼ Nsec  K  3:5;

(10.10)

The necessary spectrum bandwidths can be obtained as in Table 10.6. Note that Nsec = 1 for omnidirectional cells and Nsec = 3 for sectorial cells.

10.6.3 Hypothesis and Assumptions with Relays If the topology with relays from Chapter 9 is considered, the assumptions for costs with relays are the following: (i) Cost for BS and RSs l l l

CBS-omni = 9,000 € CRS =9,000/5 = 1,800 € CBS-trisect = 15,000 €

In these cells there are three coverage areas with an area and the BS plus three RS need to guarantee the coverage for the whole area of cell. The equivalent cost for the “BS” meaning an average between the BS and the RS is given by: CBSequivalent ¼

ðCBS þ 3CRS Þ 3

(10.11)

410

F.J. Velez et al.

ii) Cost for backhaul The cost for backhaul is the same as in case of without relays for RSs and BS that is: Cbhequivalent ¼ 1=3Cbh

(10.12)

for each hexagonal coverage area (as backhaul is only needed for the central coverage area of the BS). iii) Installation Cost The cost for installation is the same for every BS and RS. It is four times the installation cost of a BS, Cinst. Hence, we need to multiply Cinst by 1/3 to obtain the installation cost for each coverage area. It is given by: Cinstequivalent ¼ 4=3Cinst

(10.13)

iv) Maintenance and operation cost We assume that the maintenance and operation cost for the BS are the same as in the case without relays and the ones for the RS are ½. It is given by

CM&Oequivalent ¼

  CM&O þ 3=2CM&O 3

:

(10.14)

These equations may be applied to the topology with relays and omnidirectional BSs. In this case the following parameters were used, Table 10.7: l l l l

CBS-omni = 9,000 € and CRS = 1,800 € (i.e., CBS-equivalent = 4,800 €) Cinst = 1,000 € (i.e., Cinst-equivalent = 1,333.33 €) Cbh = 2,500 (i.e., Cbh-equivalent = 833.33 €) CM&O = 1,000 € (i.e., CM&O-equivalent833.33 €)

For the tri-sectored BS antennas the parameters are the following, Table 10.7 (note that the costs for the backhaul, and maintenance and operation are the same): l l l l

CBS-tri-sect = 15,000 € and CRS = 1,800 € (i.e., CBS-equivalent = 6,800 €) Cinst = 1,500 € (i.e., Cinst-equivalent = 2,000 €) Cbh = 2,500 (i.e., Cbh-equivalent = 833.33 €) CM&O = 1,000 € (i.e., CM&O-equivalent833.33 €) Table 10.7 Costs with relays for omnidirectional (K = 3) and tri-sectored (K = 1) BS antennas Costs Omnidirectional Tri-sectored K=3 K=1 47.14 47.14 Cfi(€/km2) 4,800 6,800 CBS(€) 1,333.33 2,000 CInst(€) 833.33 833.33 Cbh(€) 833.33 833.33 CM&O(€/year)

10

Business Models and Cost/Revenue Optimization

411

It should be noted that l

l

With sectorization the cost for the licence with K = 1 are equal to the cost for the licence with omnidirectional BS antenna and K = 3. With tri-sectorization some costs are higher (e.g., BS and its installation). However, higher costs are compensated with the higher revenues. Because throughput is higher (mainly due to the difference on the sub-frame format), a 9 gain of 3=5 1=3 ¼ 5 ¼ 1:8occurs, which compensated the lowest value of the frame throughput. Besides, it is worthwhile to note that

l

l

For K = 3 with relays UL traffic is  15 times the DL traffic, hence revenues are lower; For K = 1 with relays and sectorization, UL traffic is 37  DL traffic, hence revenues may be relevant.

10.6.4 Optimization and Profit Without Relays In seeking profit optimisation, revenues should be maximised with respect to costs. Under hypothesis B, which corresponds to the lowest cost case, and for K = 7, the variation of the cost and revenue in €/km2 with R is depicted in Figs. 10.5 and 10.6, for omnidirectional and tri-sectored cells respectively. The revenue curves were obtained for R144[€/min] = 0.0025 (which approximately corresponds to the price per MByte [5]). Note that the UL curves for the cases with omnidirectional BS antennas and suchannelisation and tri-sectorization are superimposed for distances up to

Hypothesis B

10000

Cost, Revenue [ /km2]

8000 C - omni. 6000

Rv - DL - omni. Rv - UL - omni. Rv - UL - Subchannel.

4000

2000

0

0

1000

3000

2000

4000

5000

R [m]

Fig. 10.5 Cost and revenues versus R for K = 7 under hypothesis B, for the omnidirectional case

412

F.J. Velez et al.

Cost, Revenue [ /km2]

250000

Hypothesis B

200000 C - Sector. Rv - UL&DL - Sector.

150000

Rv - UL - Subch. & Sector.

100000

50000

0

0

1000

2000

3000

4000

5000

R [m]

Fig. 10.6 Cost and revenues versus R for K = 7 under hypothesis B, for the tri-sectored case

3,000 m. With tri-sectorial BS antennas, as there is three time more available resources the revenues increase significatively. In order to optimize the broadband wireless access network, it is important to analyse the profit per unit area. However, it is not sufficient to compute the absolute profit because, as is shown in Figs. 10.5 and 10.6, a certain level of profit may correspond to different values of cost. For example, cost is higher for tri-sectorial cells; hence, revenue needs to be higher to obtain the same profit. It is worthwhile to note that, with tri-sectorial BS antennas, higher revenues are expectable not only due to the interference mitigation caused by the use of directional antennas but also because three channels are now available in the cell, one for each sector, leading to an higher supported throughput per km2. We have obtained results for the profit in percentage for two values of the price per MByte R144[€/min] = 0.005 and 0.010) in hypothesis A, corresponding to higher costs, in Figs. 10.7 and 10.8 respectively. It is particularly evident that profit increases as the price per MByte increases; nevertheless, the curves keep the same shape and behaviour. For tri-sectorial cells, as there is three carriers available in the cell, the profit in percentage is almost three times higher than the one for the case of omnidirectional cells for a considerable range of coverage distances, typically lower than 1,800 m, which only reaches 375%. For coverage distances larger than this value, this relative difference is only kept if sub-channelisation is considered. In hypothesis B (lower network cost), the advantage of using a tri-sectored configuration becomes even more evident for the shortest propagation distances of up to 1,250–1,500 m (see Fig. 10.9). Although the required bandwidth in each FDD link is three times higher, the profit in percentage terms is now more than three times higher than the one with omnidirectional cells. Figure 10.9 presents an example for R144[€/min] = 0.005 in hypothesis B, where it is clear that profit values

10

Business Models and Cost/Revenue Optimization

413

Hypothesis A

600

DL - omnid. UL - omnid.

500

UL - subchannel. UL&DL - sector.

400 Profit [%]

UL - subch. & sector.

300 200 100 0 –100

0

1000

3000

2000

4000

5000

R [m]

Fig. 10.7 Profit in percentage terms versus R for K = 7 under hypothesis A, R144[€/min] = 0.005

Hypothesis A

1200

DL - omnid. UL - omnid.

1000

UL - subchannel. UL&DL - sector.

Profit [%]

800

UL - subch. & sector.

600

400

200

0

0

1000

3000

2000

4000

5000

R [m]

Fig. 10.8 Profit in percentage terms versus R for K = 7 under hypothesis A, R144[€/min] = 0.010

are more than four times the values in Fig. 10.7. As costs are lower, even with a price per MByte of only R144[€/min] = 0.005 the profit in percentage terms may exceed 1,500%. Although the curves are not presented here, even for R144[€/min] = 0.0025, the values of the profit are more than twice the ones from Fig. 10.7.

414

F.J. Velez et al. Hypothesis B

2000

DL - omnid. UL - omnid.

1600

UL - subchannel.

Profit [%]

UL&DL - sector. 1200

UL - subch. & sector.

800

400

0

0

1000

3000

2000

4000

5000

R [m]

Fig. 10.9 Profit in percentage terms versus R for K = 7 under hypothesis B, R144[€/min] = 0.005

Hypothesis B

2000

K=7, tri-sect. K=4, tri-sect. K=3, tri-sect. K=1, tri-sect. K=7, omnid.

Profit [%]

1600

1200

800

400

0

0

1000

2000

3000

4000

5000

R [m]

Fig. 10.10 Profit in percentage terms versus R for different K’s under hypothesis B, with tri-sectored cells for the UL where R144[€/min] = 0.005

A comparison between K = 7, K = 4, K = 3 and K = 1 is presented in Fig. 10.10 (for tri-sectored cells in the UL and R144[€/min] = 0.005). Sub-channelisation is considered, as it is specified on the UL. Note that in the tri-sectorial case the required bandwidth in each FDD link is three times higher. For example, for K = 3,

10

Business Models and Cost/Revenue Optimization

415

it increases from 10.5 to 31.5 MHz. The consideration of a reuse pattern K = 7 is ideal in terms of interference mitigation, although K = 1, K = 3 and K = 4 may also be interesting possibilities for coverage distances between 1,500 and 1,800 m. Note, however, that, from the spectrum regulation point of view, to use K = 7 and tri-sectored cells seems to be infeasible, as a spectral bandwidth of 73.5 MHz is needed. With K = 1 and K = 3, the use of omnidirectional configurations cannot be supported as the throughput on the outer coverage ring of the cell reaches 0 Mb/s and full cell coverage is not guaranteed. Although the curve is not presented here, in the omnidirectional case, with K = 4, profits of 260–290% are achieved in the UL (and of 370–420% in the DL) for coverage distances up to 1,500 m (R144[€/min] = 0.005). Only with K = 7 higher profits are achieved with omnidirectional cells, of the order of 400–490% in the UL, as shown in Fig. 10.10 (where sub-cahnnelisation is considered), and 470–580% in the DL. However, this compares poorly with the values varying from 1,400% to 1,900% for K = 3, 4 and 7 with tri-sectored cells. The cases K = 1, K = 3, K = 4 (all with tri-sectored cells), and K = 7 (with omnidirectional cells) correspond to a spectrum bandwidth of 10.5, 31.5, 42.0 and 24.5 MHz, respectively. K = 4 and K = 3 with tri-sectorial BS antennas seem to be the optimum choices, with an advantage for the choice of K = 3, as only 75% of the bandwidth is needed. With K = 1 and tri-sectored cells the profit in percentage terms is higher than with K = 7 and omnidirectional cells, with an additional advantage: only 43% of the bandwidth is needed. Besides, it is worthwhile to note that, for K = 1 and coverage distances longer than 4,000 m, the values for the profit in percentage terms become comparable to ones with K = 4 and K = 3 (tri-sectored cells), that is, for sparse BS deployments in low density user environments K = 1 may be a solution. If network costs are higher (hypothesis A), the relative behaviour is basically the same as in hypothesis B. Under the price per MByte of R144[€/min] = 0.010, the profit may reach values of somewhere between 780% and 1,060%. It seems that the relative disadvantage with K = 7 and omnidirectional cells becomes less evident, as shown in Fig. 10.11 under hypothesis A (in comparison to Fig. 10.10). With K = 1 and tri-sectored cells, the profit in percentage terms is again higher than with K = 7 and omnidirectional cells. However, for the longest coverage distances, no advantage of using K = 1 can be found in comparison to K = 3 or 4 anymore. In terms of the choice of optimum coverage distance, from analysis of the results in Figs. 10.5 and 10.6 it is evident that network cost strongly increases if the coverage distance decreases, particularly for the lowest coverage distances. Investigating the profit in percentage terms, it can be observed that significant falls occur for coverage distances of higher than 1,500 m. Hence coverage distances of around 1,000 m might be chosen as optimum, as they maximize profit in percentage terms while keeping costs acceptable. A daily equivalent operation in saturated conditions over 6 h has been assumed, which is only valid if the offered traffic is high enough in this time span. Otherwise, revenue will be lower, and low costs will definitively reduce the possibility of losses.

416

F.J. Velez et al. Hypothesis A

1200

K=7, tri-sect. K=4, tri-sect. K=3, tri-sect. K=7, omnid. K=1, tri-sect.

Profit [%]

960

720

480

240

0

0

1000

2000

3000

4000

5000

R [m]

Fig. 10.11 Profit in percentage terms versus R for different K’s under hypothesis A, for the UL where R144[€/min] = 0.010

10.6.5 Optimization and Profit with Relays 10.6.5.1

DL Analysis with K = 3 and K = 1

In seeking profit optimisation, revenues should be maximised with respect to costs. By using Table 10.7 (but also Table 10.6), and the results for the supported throughput from Chapter 9, we obtained the curves for costs, revenues and profit in percentage. The costs and revenues with relays (K = 3, omnidirectional BS antenna), in €/km2, are depicted in Figs. 10.12 and 10.13, for R144 = 0.0025 and 0.005 €/min, respectively. In order to optimize the broadband wireless access network, it is important to analyse the profit per unit area. However, it is not sufficient to compute the absolute profit because, as is shown in Figs. 10.12 and 10.13, a certain level of profit may correspond to different values of cost. For example, cost is higher for tri-sectorial cells; hence, revenue needs to be higher to obtain the same profit. This justifies the need to represent the profit in percentage, as defined by (10.7).The operator’s/ services provider’s goal is to optimise this profit in percentage. Results are obtained with relays and K = 1 (tri-sectorial BS antennas) and K = 3 (omnidirectional BS antennas) in the DL case. Figure 10.14 presents the results for R144 = 0.0025 €/min while Fig. 10.15 presents the case for R144 = 0.005 €/min. the case without relays (K = 3) is also presented, for comparison purposes. It is clear that the use of relays without sectorization in the BS leads to a lower profit (K = 3). Even the use of sectorization (an example is presented for K = 1) does not enable to achieve larger profit (note that for K = 1 interference poses an extra

10

Business Models and Cost/Revenue Optimization

417

6000 Cost, Revenue [ /km2]

With relays, K = 3 (0.0025 /min) 5000 4000 Rv - DL

3000

C -Omni

2000 1000 0

0

1000

2000

3000 R [m]

4000

5000

6000

Fig. 10.12 Cost and revenues with relays (K = 3, omnidirectional BS antenna) for a price per MB R144 = 0.0025 €/min, in the DL

6000

Cost, Revenue [ /km2]

With relays, K = 3 (0.005 /min) 5000 4000 Rv - DL

3000

C-Omni 2000 1000 0

0

1000

2000

3000

4000

5000

6000

R [m]

Fig. 10.13 Cost and revenues with relays (K = 3, omnidirectional BS antenna) for a price per MB R144 = 0.005 €/min, in the DL

limitations; hence, results with K = 3 may be checked with sectorization). The optimum (maximum) values occur for coverage distances up to 1,000 m.

10.6.5.2

UL Analysis with K = 3 and K = 1

We also have performed analysis for UL with K = 3 and K = 1. By using the Table 10.7 (but also Table 10.6), and the results for the supported throughput from Chapter 5, we obtained the curves for the costs, revenues and profit in percentage.

418

F.J. Velez et al. 200 DL, with relays (K= 3)

Profit [%]

150

DL, no relays (K= 3) DL, sect. & relays (K= 1)

100

50

0

–50

0

1000

2000

3000

4000

5000

R [m]

Fig. 10.14 Profit in percentage for a price per MB R144 = 0.0025 €/min, in the DL 500 DL, with relays (K= 3) DL, no relays (K= 3)

400 Profit [%]

DL, sect. & relays (K= 1) 300 200 100 0

0

1000

2000

R [m]

3000

4000

5000

Fig. 10.15 Profit in percentage for a price per MB R144 = 0.005 €/min, in the DL

The costs and revenues with relays (K = 3, omnidirectional BS antenna), in €/km2, are depicted in Figs. 10.16 and 10.17, for R144 = 0.0025 and 0.005 €/min, respectively. It is shown that, in the UL, for K = 3 and omnidirectional BS antennas there is no profit, that is, revenues are always lower than costs. Results for the profit in percentage, as defined by Eq. (10.7), are obtained with relays and K = 1 (tri-sectorial BS antennas) and K = 3 (omnidirectional BS antennas) in the UL case. Figure 10.18 presents the results for R144 = 0.0025 €/min while Fig. 10.19 presents the case for R144 = 0.005 €/min. The case without relays (K = 3) is not presented, as it is impossible to obtain results with such low reuse pattern with no relays (and no sectorization, too). In the UL, with relays, only the use of sectorization (an example is presented for K = 1) enables to achieve positive profit in percentage only for R144 = 0.005 €/min

10

Business Models and Cost/Revenue Optimization

419

5000

Cost, Revenue [ /km2]

With relays, K = 3 (0.0025 /min) 4000 Rv - DL

3000

C-Omni 2000 1000 0

0

1000

2000

3000

4000

5000

6000

R [m]

Fig. 10.16 Cost and revenues with relays (K = 3, omnidirectional BS antenna) for a price per MB R144 = 0.0025 €/min, in the UL

5000 Cost, Revenue [ /km2]

With relays, K = 3 (0.005 /min) 4000 Rv - DL

3000

C-Omni 2000 1000 0

0

1000

2000

R [m]

3000

4000

5000

Fig. 10.17 Cost and revenues with relays (K = 3, omnidirectional BS antenna) for a price per MB R144 = 0.005 €/min, in the UL

(nor for R144 = 0.0025 €/min). With omnidirectional BS antennas (an example is presented for K = 3) a positive profit is not achievable in any of the prices. As for the DL, the maximum values occur for coverage distances up to 1,000 m.

10.7

Conclusions

Appropriate business models are essential to establish the business case for WiMAX networks. This Chapter started by covering general aspects about the business models for WiMAX and then addressed the cost/revenue optimization

420

F.J. Velez et al. 40 UL, with relays (K= 3)

20

UL, sect. & relays (K= 1)

Profit [%]

0 –20 –40 –60 –80 –100

0

1000

2000

3000

4000

5000

R [m]

Fig. 10.18 Profit in percentage for a price per MB R144 = 0.0025 €/min, in the UL

200 UL, with relays (K= 3)

150

UL, sect. & relays (K= 1)

Profit [%]

100 50 0 –50 –100

0

1000

3000

2000

4000

5000

R [m]

Fig. 10.19 Profit in percentage for a price per MB R144 = 0.005 €/min, in the UL

for these networks, for cellular configurations without and with relays. From an undertaken cost-benefit analysis, one conclusion of this work is that given today’s hypothesis of price per MByte of information transfer of somewhere between 0.0025 € and 0.010 €, it is clear that, without considering the use of relays, the choice of reuse patterns 3 or 4 with sectorial cells is preferable to the use of omnidirectional cells with reuse pattern of 7, as three times more resources are available in each cell. Besides, in nowadays networks, if there is a need for sparse BS deployments whilst reducing costs, K = 1 may be a solution, as it presents higher profit for the longest coverage distances. In future networks, when costs will be lower, the advantage of sectorization is kept and will drive the deployment of

10

Business Models and Cost/Revenue Optimization

421

tri-sectorization forward. Nevertheless, in this case K = 1 will not be advantageous with tri-sectorization for the longest coverage distances anymore. It has also been concluded in this paper, driven by our analysis as well as other observations, that cell radii in the range 1,000–1,500 m might be chosen. This range corresponds to a profit, in percentage terms, of near to the maximum achievable, while keeping costs acceptable. The WiMAX cost-benefit optimization is also explored for the case where relays are used to help on improving coverage while mitigating the interference. Results show that the use of relays with no sectorization in the BS leads to a lower profit (K = 3). Also the use of sectorization (an example is presented for K = 1) does not enable to achieve larger profit. The optimum (maximum) values occur for coverage distances up to 1,000 m. In the DL, when the price per MB, R144, increases from 0.0025 €/min to 0.005 €/min the profit increases more than 100%. Suggestion for future work on the optimization of cellular configurations with relays is to explore sectorization for K = 3 (as interference is mitigated for larger reuse patterns), consider different assumptions for the prices of the relays, and mainly to jointly achieve the profit curves joining together the UL and DL contributions for the supported traffic.

References 1. H. Chesbrough, R.S. Rosenbloom, The role of the business model in capturing value from innovation: evidence from Xerox Corporation’s technology spin-off companies. Ind. Corporate Change 11(3) (June 2002) 2. WiMAX Forum whitepaper, WiMAX Technology Forecast (2007–2012) (June 2008) 3. M. Ibrahim, K. Khawam, A.E. Samhat, S. Tohme, Analytical framework for dimensioning hierarchical WiMax-WiFi networks. Comput. Netw. 53(3), 299–309 (2009) 4. B. Gavish, S. Sridhar, Economic aspects of configuring cellular networks. Wireless Netw. 1(1), 115–128 (Feb 1995) 5. F.J. Velez, L.M. Correia, Optimisation of mobile broadband multi-service systems based in economics aspects. Wireless Netw. 9(5), 525–533 (Sept 2003) 6. K. Johansson, A. Furusk€ar, P. Karlsson, J. Zander, Relation between cost structure and base station characteristics in cellular systems, in Proceedings of PIMRC’ 2004 – 15th IEEE International Symposium on Personal, indoor and Mobile Radio Communications, Barcelona, Spain, Sept 2004 7. D. Reed, The cost structure of personal communications, IEEE Commun. Mag. 31(4), 102–108 (Apr 1993) 8. J. Sarnecki et al., Microcell design principles, IEEE Commun. Mag. 31(4), 76–82 (Apr 1993) 9. F.J. Velez, A.H. Aghvami, O. Holland, Basic limits for fixed WiMAX optimization based in economic aspects, accepted for publication in IET Communications – Special Issue on WiMAX Integrated Communications (Mar 2009) 10. F.J. Velez, L.M. Correia, Cost/revenue optimisation in multi-service mobile broadband systems, in Proceedings of PIMRC’ 2002-13th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications, Lisbon, Portugal, Sept 2002

Chapter 11

Multiple Antenna Technology Ramjee Prasad and Fernando J. Velez

Abstract As the radio resources are scarce and data rate requirements keep increasing, spectral efficiency is a stringent requirement in present and future wireless communications systems, MIMO-OFDM has become a new “star” in the “constellation” of wireless and mobile communications. Its potential to increase spectral efficiency has not been reached by any other technology before. In addition to increasing spectral efficiency, MIMO can also be used to reduce transmitting power while keeping coverage areas constant. The use of MIMO technology in future transmission systems for broadcasting, multicasting and unicasting represents real business logic also for broadcasting corporations because of the possible reduction in transmission stations. This Chapter covers Closed Loop MIMO, multiantenna in OFDM, space diversity, space-time coding, Allamouti code, spatial multiplexing, collaborative spatial multiplexing, diversity and array gains, system model for MIMO OFDM, support of MIMO in IEEE 802.16, beamforming and trade-off between spatial multiplexing and spatial diversity, among other issues.

11.1

Introduction to MIMO

The major performance degradation factors for the mobile wireless channel are channel fading and co-channel interference. These degradation factors are better handled by using advanced antenna technologies. Advanced antenna technologies employ multiple antennas at the transmitter as well as at the receiver side to achieve more reliable and higher data rates. To put it in simple terms, Multiple Input Multiple Output (MIMO) can be used to refer any multiple antenna technologies.

R. Prasad (*) Center for TeleInFrastruktur (CTIF), Aalborg University, Niels Jernes Vej 12, DK–9220 Aalborg Øst, Denmark e-mail: [email protected]

R. Prasad and F.J. Velez, WiMAX Networks, DOI 10.1007/978-90-481-8752-2_11, # Springer ScienceþBusiness Media B.V. 2010

423

424

R. Prasad and F.J. Velez

MIMO Multiple Input Multiple Output

Open Loop MIMO

Matrix A Space Time Block Coding

Matrix B Spatial Multiplexing

Closed loop MIMO

Beamforcing Transmitter Adaptive Antenna

Fig. 11.1 WiMAX multi-antenna implementation chart

In application, the MIMO technology can be divided based on Open Loop MIMO techniques and Closed Loop MIMO techniques. MIMO systems that do not rely on knowledge of the channel responses at the transmitter are open loop. Closed loop is the case when the channel state information (CSI) is available at the transmitter through some form of feedback mechanism. Regarding MIMO options, the WiMAX standard includes two versions of Open Loop MIMO techniques, for the use on the downlink referred to as MIMO Matrix A and MIMO Matrix B. IEEE 802.16e supports both the techniques and therefore form a part of wave 2 certification. Closed Loop MIMO techniques are also called Transmitter Adaptive Antenna techniques or “beamforming” (Fig. 11.1). l l l

Space Diversity Spatial Multiplexing Beamforming

IEEE 802.16 uses a radio modulations scheme called Orthogonal Frequency Division Multiplexing (OFDM). The important characteristic of OFDM with regards to SISO/MIMO is that OFDM divides data frames into numerous smaller symbols for transmission. In SISO, each of these symbols is transmitted serially, one after the other. WiMAX is based on smart antenna friendly OFDM/OFDMA technology. OFDM/OFDMA is a powerful means to convert a frequency selective wideband channel into multiple flat narrow and subcarriers. This allows smart antenna operations to be performed on vector flat sub-carriers. The structure of the remaining of this Chapter is as follows. After presenting the MIMO wireless transmission system and the closed loop MIMO in this Section, Section 11.2 addresses the capacity of MIMO channels. Section 11.3 presents multi-antenna in OFDM. Sections 11.4 and 11.5 present spaced diversity and space-time coding. Section 11.6 addresses spatial multiplexing while Section 11.7

11

Multiple Antenna Technology

425

Mt SpaceSpaceTime Mapper transmitter

...

Enc

Mr

...

binary I/P

Constel. mapping Constel. mapping

channel

SpaceSpaceTime detection detection & decoding decoding &

binary O/P

receiver

Fig. 11.2 MIMO wireless transmission system

describes collaborative spatial multiplexing. Section 11.8 presents the diversity gain and array gain while Section 11.9 addresses the comparison of MIMO options. Section 11.10 presents the MIMO model for MIMO OFDM. The support of MIMO technology in the IEEE 802.16 standard is introduced in Section 11.11 while Section 11.12 addresses beamforming. Section 11.13 covers the trade-off between spatial multiplexing and spatial diversity. Finally, conclusions are drawn in Section 11.14.

11.1.1 MIMO Wireless Transmission System Figure 11.2 shows basic MIMO systems can be defined simply. Both the transmitting and the receiving ends are equipped with multiple antenna elements. In MIMO the signals on the transmitter (TX) antennas, at one end, and the receiver (RX) antennas, at the other end, are “combined” in such a way that the quality of the communications in terms of bit-error rate (BER) or data rate (bit/s) for each MIMO user will be improved. Such an advantage can be used to increase significantly both network’s QoS and operator’s revenues. As shown in the Fig. 11.2 input, in the form of a binary data stream, is fed to a simplified transmitting block encompassing the functions of error control coding and mapping to complex modulation symbols, for example, quaternary phase-shift keying (QPSK) or quadrature amplitude modulation with M symbols (M-QAM). Several separate symbol streams are produced. They range from independent to partially or fully redundant. Each stream is then mapped onto one of the multiple TX antennas. The signals are launched in the wireless channel after upward frequency conversion, filtering and amplification. At the receiver side the string of Mr multiple antennas capture the signals and demodulation and demapping operations are performed to recover the original message. MIMO takes advantage of random fading [1–3] and, when available, multipath delay spread [4, 5], for multiplying transfer rates. A key feature of MIMO systems is the ability to turn multipath propagation, traditionally a pitfall of wireless transmission, into a benefit for the user. The goal of MIMO is to improve the spectral efficiency (bit/s/Hz), the coverage area (cell radius) and the signal quality (bit-error rate or packet-error rate). MIMO systems are becoming popular and find and are applied for emerging wireless technologies, such as WLAN,

426

R. Prasad and F.J. Velez

Broadband Wireless Access (BWA) and cellular. These advances do however come at additional cost. Multiple antennas increase RF costs and complexities and mathematically complex DSP algorithms challenge the designers and manufacturers. Using MIMO communication system at the transmitter and the receiver end gives the following advantages: l l l l

Increase the coverage area Increase the achievable data rate Decrease in BER, hence increase the system reliability Decreases the required transmit power

11.1.2 Closed Loop MIMO When the transmitter can acquire some sort of knowledge of the channel between it and the receiver, it can adjust the transmitted signal accordingly in order to optimize the system performance or the data rate. As the channel changes quickly in a highly mobile scenario closed loop transmission scheme are primarily feasible in fixed or low mobility scenarios. Channel state information (CSI) plays a critical role in achieving reliable and high rate communications. In closed-loop spatial multiplexing, every stream is transmitted from all of the antennas using weights computed from the channel estimation, as shown in Fig. 11.3. The CSI can be used to weight the signals transmitted from the BS antennas. The weighting should be such that the signals arrive co-phased in the receiver. Constructive signal combining increases the received signal power, therefore SNR gain and link capacity will also increase. Figure 11.4 presents a comparison between closed and open-loop MIMO. H s1 s2 . . .

Precoding

Rx

W

sm Nr

Nt

Feedback channel

Fig. 11.3 Closed loop transmit diversity

11

Multiple Antenna Technology

a

427 b

1x1 vs. 2x2 vs. 2x3, MMSE, 40 MHz, model D, NLOS

200

1x1 open loop 2x 2 open loop 2x3 open loop 2x 2 closed loop

20 MHz, convolutional code, model D NLOS 220 200 180

Throughput (Mbps)

Throughput (Mbps)

250

150

100

1x1, 20 MHz 4x4, 20 MHz, MMSE 4x4, 20 MHz, SVD 4x4, 20 MHz, SVD with puncturing

160 140 120 100 80 60

50

40 20

0 –10

0 –5

0

5 10 15 SNR = Es/N0 (dB)

20

25

30

5

10

15

20

25

30

SNR = Es/N0 (dB)

Fig. 11.4 Comparison between closed-loop and open-loop MIMO: (a) 2  2 closed-loop equals 2  3 open-loop, and (b) 4  4 closed-loop 5 dB better than 4  4 open-loop

11.2

Capacity of MIMO Channel

In simple terms channel capacity refers to the maximum data rate. The sent signal is represented by s and the received signal is r [6]. A time-invariant and narrowband channel is assumed. The channel model includes the matrix H with the direct and the indirect channel components (Fig. 11.5). The received signal can be computed as follows r ¼ Hs þ n;

11.3

(11.1)

Multi-antenna in OFDM

The main motivation for using OFDM in a MIMO channel is because OFDM modulation turns a frequency-selective MIMO channel into a set of parallel frequency-at MIMO channels. This makes multi-channel equalization particularly simple, since for each OFDM-tone only a constant matrix has to be inverted [6, 7]. In frequency selective environments, however, the amalgamation of Spatial Multiplexing (SM) and OFDM techniques can be a potential source of high at reasonable complexity. Figure 11.6 shows a schematic of a MIMO-OFDM system [7]. OFDM Modulation (OMOD) and OFDM Demodulation (ODEMOD) denote an OFDM-modulator and demodulator, respectively. The basic principle that underlies OFDM is the insertion of a guard interval, called cyclic prefix (CP), which is a copy of the last part of the OFDM symbol, and has to be long enough to accommodate the delay spread of the channel. The use of the CP turns the action of the channel on the

428

R. Prasad and F.J. Velez h11 S1

Space-Time encoder

h12

S2

h21 hMr 1

r1

h22

r2

hMr 2

h1Mt SMt s

Space-Time decoder

h2Mt

r1

r Mr

hMr Mt H

r

Fig. 11.5 Physical MIMO channel

OMOD

ODEMOD and separation

OMOD

OMOD

Fig. 11.6 Schematic of a MIMO-OFDM system (adapted from [7])

transmitted signal from a linear convolution into a cyclic convolution, so that the resulting overall transfer function can be diagonalized through the use of an IFFT at the transmitter and an FFT at the receiver. Hence, the overall frequency-selective channel is converted into a set of parallel flat fading channels, which drastically simplifies the equalization task. Let us assume an OFDM-MIMO system with Nt transmit antennas, Nr receive antennas and N subcarriers. The input-output relation for the MIMO system for the kth tone, k = 1,. . ., N may be expressed as rffiffiffiffiffiffi PT Yk ¼ H k sk þ N k ; NT

(11.2)

where Yk and Nk are NR  1, Hk is a NR  NT matrix, sk is a NT  1 vector, and PT is the total transmit power. The matrix Hk is the frequency response of the matrix channel corresponding to the kth tone.

11

Multiple Antenna Technology

429

From the above equation we can see that, just as in SISO channels, OFDMMIMO decomposes the otherwise frequency selective channel of bandwidth 5 into N orthogonal flat fading MIMO channels, each with bandwidth B = N. MIMO signalling treats each OFDM tone as an independent narrowband frequency flat channel [8]. We must take care to ensure that the modulation and demodulation parameters (carrier, phasing, FFT/IFFT, prefixes, etc.) are completely synchronized across all transmit and receive antennas. With this precaution, every OFDM tone can be treated as a MIMO channel, and the tone index can be treated as a time index in the already existing Space-Time (ST) techniques.

11.4

Space Diversity

High degree of attenuation in a multipath wireless environment makes it extremely difficult for the receiver to determine the transmitted signal. This problem is addressed by transmitting less-attenuated replica of the original signal to the receiver, that is providing with some form of diversity. Diversity is proven to be an effective tool for coping with the challenges of NLoS propagation. Multi-path and reflections signals that occur in NLoS conditions may be effectively reduced by using diversity schemes. IEEE 802.16e uses transmit diversity in the downlink direction to provide spatial diversity that enhance the signal quality to a specific subscriber located anywhere within the range of the antenna beam. Receive diversity is obtained by various combining techniques to improve the availability of the system. For instance, maximum ratio combining (MRC) takes advantage of two separate receive chains to help in overcoming fading and reducing path loss [9]. On the one hand, transmit diversity provides less signal gain compared to beamforming. On the other, transmit diversity is more robust for mobile users since it does not require prior knowledge of the path characteristics for a subscriber’s particular frequency channel.

11.5

Space-Time Coding

Space-time coding (STC) is a family of techniques for implementing transmission diversity. The diversity gain is obtained by applying coding techniques across space and time. Among the various space time diversity codes, it is worthwhile to briefly describe the space-time-trellis code (STTC) and space-time block code (STBC). In its original form, STTC is a two dimensional trellis code and its complexity increases exponentially with the number of transmit antennas and the constellation

430

R. Prasad and F.J. Velez

order. Another popular form of STC technique is STTC Alamouti Code [9], which is defined for two transmit antennas. An STC is represented by a matrix. Each row of the matrix denotes a time slot while each column represents the different antennas transmissions over time. S11 S21  ST1

S12 S22  ST2

   

S1nT S2nT  STnT

(11.3)

The code rate of an STBC measures how many symbols per time slot (T represents number of time slot) it transmits on an average over the course of a block of size k. Thus, the code rate, r, can be obtained as follows r ¼ k=T ;

(11.4)

11.5.1 Alamouti Code Among all the STCs defined for different number of transmit antennas, Alamouti Code is the simplest. The following matrix illustrates the STC Matrix A scheme 

S A¼ 1 S2

S2 S1

 (11.5)

Two information symbols S1 and S2 are transmitted over a period of 2 symbols and sent using a specific coding between the two antennas. Figure 11.7 shows the

h1 s1,s2

r1

s1 s2*

h2 s2 – s1* r1 = h1s1+ h2s2+ n1 r2 = h 1 s 2*– h2 s1*+ n2

Fig. 11.7 Matrix A Alamouti’s STC

r2

11

Multiple Antenna Technology

431

space-time code of the 2  1 Alamouti Code solution, also called STC matrix A scheme. The first row corresponds to the symbols transmitted from the first antenna while the second one corresponds to the symbols transmitted from the second antenna. The first column represents the first transmission period while the second column the second transmission period. Elaborating further, during the first symbol period, the first antenna transmits S1 and the second antenna transmits S2. During the second symbol period, the first antenna transmits S1 and the second antenna transmits S2 , the complex conjugate of S1 and S2, respectively. This implies that we are transmitting both in space (across two antennas) and time (two transmission intervals). This is space-time coding. The signals received for two symbol periods are represented by, Suppose that (s1, s2) represent a group of two consecutive symbols in the input data stream to be transmitted. During a first symbol period t1, transmit (Tx) antenna 1 transmits symbol s1 and Tx antenna 2 transmits symbol s2. Next, during the second symbol period t2, Tx antenna 1 transmits symbol s2 and Tx antenna 2 transmits symbol  s1 . Denoting the channel response (at the subcarrier frequency at hand) from Tx1 to the receiver (Rx) by h1 and the channel response from Tx2 to the receiver by h2, the received signal samples corresponding to the symbol periods t1 and t2 can be written as 

r1 ¼ h1 s1 þ h2 s2 þ n1 ; r2 ¼ h1 s2 þ h2 s1 þ n2

(11.6)

where, n1 and n2 are samples of white Gaussian noise. Here, one denotes the channel response from Tx antenna 1 to the receiver by h1 and its response from Tx antenna 2 to the receiver by h2. Assuming a flat-fading channel, h1 is the complex channel gain from antenna 1 to the receive antenna, and h2 is from antenna 2, the received signal samples at the two transmission instants are given by (11.6). In matrix A, S1 and S2 are the symbols to be transmitted over the air. At the symbol k, symbol S1 is transmitted from antenna 1 while symbol S2 is transmitted from antenna 2. At the symbol k + 1, symbol –S1* is transmitted from the antenna 2. The receiver computes the following signals to estimate the symbols s1 and s2   x1 ¼ h1  r1  h2 r2  ¼ jh1 2 þ jh2 j2 s1 þ h1 n1  h2 n2   x2 ¼ h2  r1 þ h1 r2  ¼ jh1 2 þ jh2 j2 s1 þ h1 n1  h1 n2

(11.7)

These expressions clearly show that x1, x2 can be sent to a threshold detector to estimate symbol s1 and symbol s2 respectively without interference from the other symbol. Moreover, since the useful signal coefficient is the sum of the squared modules of two independent fading channels, these estimations benefit from perfect second-order diversity, equivalent to that of Rx diversity under Maximum-Ratio Combining (MRC).

432

R. Prasad and F.J. Velez

Alamouti’s transmit diversity can also be combined with MRC when two antennas are used at the receiver. In this scheme, the received signal samples that correspond to the symbol periods t1 and t2 can be written as 

r11 ¼ h11 s1 þ h12 s2 þ n11 ; r12 ¼ h11 s2  h12 s1 þ n12

(11.8)

11.5.2 Second-Order STC Second-order STC in DL supports coding rates 1 and 2 by using the following two transmission format matrices. Here Sk’s are OFDM symbols in the frequency domain right before IFFT operation. The optional STC transmit diversity is also supported in UL using the transmission format matrix A of Eq. (11.4). Matrix B of Eq. (11.5) can be used by two SSs in a collaborative special multiplexing mode. MIMO Matrix A using STC improves the reliability of data transmission for mobile WiMAX using multiple transmit antennas. By adding redundant, parallel paths, the modem has twice as much chance of receiving a good copy of the data. Under stationary conditions, the gain provided by STC is only +3dB, but in a fading environment, such as when passing rapidly between buildings, the gain may be as much as +5dB for 16 QAM and +10dB for 64 QAM compared to a non-STC signal under the same conditions [11]. As a consequence, the system can hold a relatively high throughput under difficult conditions.

11.6

Spatial Multiplexing

Spatial Multiplexing increases spectral efficiency by transmitting multiple data streams simultaneously by using the same frequency channel at the same time by employing multiple transmit and multiple receive antennas [12]. With this